I'fH;!

sIMmM*

Handle with
EXTREME CARE
This volume is

BRITTLE
and cannot be repaired
Photoconv onlv if necessary
GERSTEIN SCIENCE
INFORMATION CENTRE
L. -
o
 ( )1
'"
. !

Practical Astronomy

FOR SURVEYORS

THE ELEMENTS OF GEODESY

LlEUT.-COLONEL J, R, OLIVER, R,A

Professor of Surveying at the Royal Military College of Canada.

KINGSTON :

PRINTED AT THE DAILY NEWS OFFICE.

1883.
 QB

Ok

1020268
 PREFACE,

This manual has been drawn up for the use of the

Cadets of the Royal Military College of Canada. The
first five chapters on Practical Astronomy embrace that
portion of the subject with which allLand Surveyors in
this country ought to be familiar. The remaining chap-
ters, together with the part of the work which treats of

Geodesy, touch on the more important parts of the ad-

ditional course, as regards those subjects, laid down by
Government for candidates for the degree of Dominion
Topographical Surveyor. It has become absolutely
necessary to draw up some compilation of this kind, be-

cause, while many of the Cadets are anxious to qualify

themselves as far as possible in the above-mentioned
course, the number of different books they would have
had to refer to in order to obtain the requisite knowledge
would have entailed on them a heavy expense. In order
to make the work as cheap as possible the number of
diagrams has been cut down to a minimum, it being in-
tended to supply the place of expensive plates of instru-
ments et cetera by lecture illustrations. The author has
also made the higher portion of the Astronomical course
iv Preface.

as brief as possible. It will be found treated in the

fullest manner in Chauvenet's Astronomy.

Geodesy being both a difficult and a very extensive sub-

ject no attempt has been made to write anything like a
treatise on it. All that has been aimed at has been to

give a sketchy account of its most salient points, adding

a few details here and there. The student who wishes to

pursue the subject further is referred to standard works,

such as Clarke's Geodesy.

The author has to acknowledge having made more or

less use of the following:

Chauvenet's Astronomy, Puissant's Ge'ode'sie, Clarke's

Geodesy, Frome's Trigonometrical Surveying, Loomis'

Practical Astronomy, Gillespie's Higher Surveying,
Deville's Examples of Astronomic and Geodetic Calcula-
tions, the U. S. Naval Text Book on Surveying, and

Jeffers' Nautical Surveying. He has also to thank Lieut. -

Colonel Kensington, R.A., for valuable assistance in in-

vestigating some doubtful formulas.

Kingston, Cadada,
January, 1883.
 CONTENTS,

PART I.

PRACTICAL ASTRONOMY.
CHAPTER I.

PAGE.
General view of the universe. The fixed stars. Their classification,
magnitudes, and distances. The sun. The planets. Their rela-
tive sizes and distances from the sun. Apparent motions of the
heavenly bodies. Their real motions. Motion of the earth with
reference to the sun. The solar and sidereal day. Mean and
apparent solar time. The equation of time. Sidereal time.
The sidereal clock i

CHAPTER II.

Idea of the great sphere. Meaning of the terms "pole," "meridian,"

"declination circle," "hour circle," "zenith," "latitude," "longi-
tude." "declination," "right ascension," "altitude," "azimuth,"
"sensible horizon," "rational horizon," "parallels of latitude,"
"declination parallels," "circumpolar star," "transit," "paral-
lax." Refraction. The Nautical Almanac. Sidereal time. The
celestial globe. Illustration of the different co-ordinates on the
great sphere 1 1

CHAPTER III.
Uses of practical astronomy to the surveyor. Instruments employed
in the field. Their particular uses. Corrections to be applied to
an observed altitude. Cause of the equation of time. Given the
sidereal time at a certain instant to find the mean time. To find
the mean time at which a given star will be on the meridian.
Given the local mean time at any instant to find the sidereal time,
Illustrations of sidereal time. To find the hour angle of a given
star at a given meridian. To find the mean time by equal alti-
tudes of a fixed star, To find the local mean time by an observed
altitude of a heavenly body. To find the time by a meridian
transit of a heavenly body 22
 vi Contents.

CHAPTER IV.
To find the latitude by the meridian altitude of the sun or a star.
The longitude. Differences of longitude measured by differences
of local time. The meridian. To find the azimuth of a
heavenly
body from its observed altitude. To find the meridian by equal
altitudes of a star. To find the meridian by the greatest elonga-
tion of a circumpolar star. To find the meridian by observations
of high and low stars. Azimuth by observations of the
pole
star at any hour
36
CHAPTER V.
Sundials. Horizontal dials. Vertical dials 49
CHAPTER VI,
The Refracting Telescope. The The Reading Micro-
Micrometer.
scope. The Spirit Level. The Chronometer. The Electro-
Chronograph. The Sextant. The Simple Reflecting Circle.
The Repeating Reflecting Circle. The Prismatic Reflecting
Circle 3
CHAPTER VII.
The portable Transit Telescope. Its uses and adjustments. Methods
of correcting the meridian line. Effect of inequality of pivots.
To apply the level correction to an observation. To find the
latitude by transits of stars across the prime vertical. Effect of
an error of deviation on the latitude. The personal equation.. 66
CHAPTER VIII.
The Zenith Telescope. Its use in finding the latitude. To find the
corrected latitude. To find the level correction. Value of a
division of the level. Value of a revolution of the micrometer
screw, Reduction to the meridian. The portable transit instru-
ment as a zenith telescope 77
CHAPTER IX.
Additional methods of finding the latitude. By a single altitude
.

taken at a known time. By observations of the pole star out of

the meridian. By circum-meridian altitudes 84
CHAPTER X.
Interpolation, by second differences. Examples. To find the Green-
wich mean time corresponding to a given right ascension of the
moon on a given day. Interpolation by differences of any order.
To find the longitude by moon-culminating stars. To find the
longitude by lunar distances 89
CHAPTER XI.
To find the amplitude and hour angle of a given heavenly body when
on the horizon. To find the equatorial horizontal parallax of a
heavenly body at a given distance from the centre of the earth.
To find the parallax in altitude, the earth being regarded as a
sphere. Star catalogues. Differential variations of co-ordinates.
Correction for small inequalities in the altitudes when finding
the time by equal altitudes. Effect of errors in the data upon
\ the time computed from an altitude. Kffect of errors of zenith
distance, declination, and time upon the latitude found by cir-
cum-meridian altitudes. The probable error 101
 Contents. vit

PART II

GEODESY.
CHAPTER I.

The figure of the garth an oblate spheroid proved by measurements

of arcs of a meridian at different latitudes. Local abnormal
deviations of the plumb line. The compression and ellipticity of
the earth. Meaning of certain terms. To find the reduction of
the latitude. To find the radius of the terrestial spheroid for a
given latitude. To find the length of the great normal for a
given latitude. To find the radius of curvature of the terrestrial
meridian for a given latitude in
CHAPTER II.

Geodetical operations. Methods adopted for mapping country. Tri-

angulation. Primary, secondary, and tertiary triangles. Base
lines. Bases of verification. The Lapland, Peru, Hounslow
Heath, French, and Lough Foyle bases. Apparatus used in
measuring. Colby's rods. The American arrangement. Piue
rods. Preliminary measurement. Permanent marks, Reduc-
tion to the horizontal. The broken base. The base with an
angle. Reduction of a base to the sea level. Measurement of
bases by sound. Astronomical base lines 120
CHAPTER III.

Triangulation. Size of primary triangles. Choice of stations and

construction of signals. Monuments. The
heliostat and electric
light. The Spain-Algiers triangulation. Measuring the angles.
Theodolites of different sizes. Repeating and reiterating theodo-
lites. Method of repeating an angle. To reduce a measured
angle to the centre of a station. Correction for phase of signal.
To reduce an inclined angle to the horizontal plane. The spheri-
cal excess. Correcting the angles of a triangle. Calculating the
sides of the triangles. The methods employed, Delambre's and
Legendre's methods. Investigation of the best form of triangle
in geodetical operations 133
CHAPTER IV.
Determination of the geodetic latitudes, longitudes, and azimuths, of the
stations of a triangulation, taking into account the ellipticity of
the earth. Formuloe and series employed. Use of the imaginary
sphere described with radius equal to the normal of the spheroid.
Reduction of a difference of latitude on the spheroid to the cor-
responding difference of latitude on this sphere. Given the
lalitudes and longitudes of two points to find the length and
direction of the line joining them. To compute the distance
between two points, knowing their latitudes and the azimuths
of one from the other. To compute the distance, knowing
the latitude of one point, the azimuth of the line connecting it
with the other, and the difference of the longitudes of the points.
Numerical examples. Method of correcting the azimuth adopted
Viil Contents.

in the North American Boundary Survey of 1845. Deville's

method of solving certain practical problems. To find the area
of a portion of the surface of a sphere bounded by two parallels
of latitude and two meridians. To find the offsets to a parallel
of latitude 145
CHAPTER V.
Methods of delineating a spherical surface on a plane. Projections.
The orthographic, stereographic, gnomonic or central, Mercator's,
and ordinary polyconic 166

CHAPTER VI.
Trigonometrical levelling. To find the height of one station above
another, Reciprocal observations for cancelling refraction. Re-
duction to the summit of the signals. Geodetical formuloe used
for the more exact determination of heights. To find the differ-
ence of level of two stations by reciprocal zenith distances, and
by a single zenith distance. Numerical example. To find the
height of a station by the zenith distance of the sea horizon. To
find the co-efficient of terrestrial refraction by reciprocal observa-
tions of zenith distances 175
CHAPTER VII.
The use of the pendulum in determining the compression of the
earth. The effect of the spheroidal form of the earth on the force
of gravity. The latter measured by the oscillations of a pendu-
lum. The length of the pendulum may be fixed and the time of
its oscillation observed, or we
may ascertain what the length of
the pendulum must be in order that it may oscillate in a given
time. The simple and the compound pendulum. The centre of
oscillation. Clairant's Theorem. Borda's pendulum. Rater's
reversible pendulum. Its uses. Results of pendulum obser-
vations 182
 NOTE TO PAGE 52.

Bv drawing a figure it can be easily shown that, in the

case of a horizontal dial, iff is the latitude, P the hour
angle, and a the angle the corresponding hour line makes
with the meridian line, then :

sin <p
cot P tan a
or tan a =sin <p
tan P.

Similarly, in the case of a dial on a vertical wall facing

south,
tan a=cos <p
tan P.

In the latter case the angle is measured from a ver-

on the wall. The stile is, of course, set parallel

tical line
to the polar axis.

We can thus find the hour lines for each hour, for any
given latitude, by solving these equations.
 Part I.

PRACTICAL ASTRONOMY.

CHAPTER I.

GENERAL VIEW OF THE VISIBLE UNIVERSE. THE FIXED

STARS. THE SOLAR SYSTEM. APPARENT AND REAL
MOTIONS OF THE HEAVENLY BODIES. DIFFERENT
METHODS OF RECKONING TIME.
The visible universe, outside our earth, comprises the
sun, moon, planets, fixed stars, milky way, nebulae, shoot-
ing stars, and the zodiacal light, besides an occasional
comet.
The comets, shooting stars, and zodiacal light will not
be further alluded to here. The milky way and the nebulae
(white cloudy patches) when examined with powerful tele-
scopes generally resolve themselves into clusters of
separate stars; a few nebulae, however, still retaining their
cloud-like appearance.

The fixed stars, as they are called, are doubtless suns,

scattered irregularly (or more properly in clusters) through
 The Fixed Stars.

space. They are classified by astronomers into magni-

tudes, the brightest being those of the ist magnitude.
Those of the 6th magnitude are about the smallest visible
to the naked eye, those below that size being only visible
through telescopes. Although differing so much in bright-
ness, the most powerful telescope fails to show them of
any measurable size, and they all appear mere points of
light. Their brightness, as seen by us, depends, probably,
partly on their distances, partly on their size, and partly
on their natural brilliancy, while that of a few
of them varies at regular intervals. The colour of the
stars also varies, inclining in some to white, in others,
to red, blue, or green. Some stars are connected in pairs
and revolve round a common centre. It has been ascer-

tained by the spectroscope that the elements present in

the sun and stars are identical with those composing
our earth ; at least no new ones have yet been discovered.

The were grouped by the ancients into constella-

stars
tions, of which the Great Bear and Orion are instances;
and a number of the most remarkable stars received
special Arabic names, e.g., Arcturus and Aldebaran. The
stars composing a constellation are catalogued according
to their brightness, the Greek letters being used to dis-
tinguish them. Thus Aldebaran is a Tauri, and the two
stars of that constellation next in brightness are ft and y
Tauri. When the Greek alphabet is exhausted English
letters are used, and finally numbers. Thus we have
h Virginis and 51 Cephei. The stars are numbered,
not according to their brightness, but in the order of their
right ascension.

The distances of the fixed stars from the earth and from
each other are so great as to be almost beyond human
conception. It was for long believed that they could not

be measured. It was. however,

eventually found thai in
the a of souk; of them, by taking a line through space
'
 Their distance.

joining opposite points of the earth's orbit as a base, and

the star as the apex of a very acute-angled triangle, the angles
adjacent to the base could be measured and the acute angle
thus determined. The length of the base being known
gives the star's distance. To give an idea how far off the
nearest star is it may be mentioned that a ray of light
would pass round the earth (about 24,900 miles) in a

quarter of a second it takes 8 minutes to traverse the

; -J-

93 millions of miles from the sun to the earth and 3J ;

years to reach us from the star. And yet, could we be

transported to that star, we should still see all the other
familiar constellationsand stars apparently in exactly the
same positions as we see them here. So vast are the dis-
tances that the change of position of the observer would
have about as much effect on that of the stars as would
an interchange of two adjoining grains of sand on a large
table covered with them.

The
nearest star, as at present known, is a Centauri,
which 200,000 times farther off than the sun. The ap-
is

proximate distances of a few others, in terms of the num-

ber of years it takes their light to reach us, are as follows :

ftCentauri 6 years.
61 Cygni 8
Sirius 16
Procyon 16
Arcturus 16
Vega 16
Pole Star 32
About 100,000 have been catalogued altogether.
stars
The number visible with the naked eye is about 15,000.
In latitude 50 north only about 2,000 can be thus seen
at any one time.
Our sun is only one of the stars, and the latter, though
called "fixed," are in reality all moving according to the
laws of dynamics. What these motions are we cannot
tell, as we do not yet know the manner in which the
 The Planets.

masses are distributed through space. It has however

been ascertained, not only that they are slowly changing
their position with regard to each other, but that in one

part of the heavens they are getting farther apart, thus

indicating that the motion of our sun with his attendant
planets is in the direction of that part. It may be inferred

that the stars are, as a rule, the centres of planetary

systems like our own, and that possibly each system has
at least one planet in a state of development permitting
of its habitation by living creatures. Our own solar sys-
tem consists of the sun and eight planets, besides a number
of small planets revolving between the orbits of Mars and
Jupiter. The planets move in ellipses, of which the sun
is a focus. Several of them have moons or satellites, and
all, including the sun, revolve on their own axes.

The planes in which the larger planets move all nearly

coincide with that of the earth, the greatest difference being
in the case of Mercury, whose orbit is inclined to ours at
an angle of y. Looking down on the plane of the sys-
tem from its northern side, the direction of the motion of
the planets round the sun, of the satellites round the
planets, and of all its members (including the sun) round
their own axes, would be seen to be opposite to that of
the hands of a clock. The earth has one moon (distant
from it about 60 radii of the earth), Mars two, Jupiter
four, Saturn eight, Uranus four, and Neptune one. The
distances of the planets from the sun are nearly in the
following proportion :
Mercury 1, Venus 2, the Earth
2.6, Mars 4, Jupiter 13, Saturn 25, Uranus 50, Neptune 80.
The density or specific gravity of Mercury is a little more,
and that of Venus a little less, than that of the earth,

of the moon fths, of the sun and Jupiter one quarter, of

Mars Wth, Saturn -th, Uranus -^th, and Neptune |th.
To compare if we took a globe four
their relative sizes;
feet indiameter to represent the sun, the moon would be
about the size of a grain of shot, Mercury of a buckshot,
 Their relative size.

Venus and theEartha small spherical rirlebullet, Mars a small

revolver bullet, Jupiter an 18-pounder round shot, Saturn
a 9-pounder round shot, and Uranus and Neptune large
grape shot; the latter the largest. The mass of Jupiter
is 300 times that of the Earth, while that of
Mercury is
only about TVth, Mars -^th, and the Moon -^th.

The planets are easily recognized by their changing

their places in the sky relatively to the fixed stars hence
their name, which means "wanderer." They may also be
known by their shining with a steady fixed light instead
of twinkling. They shine^onlyby the sunlight reflected
from their surfaces, and when viewed through a good
telescope, look like small moons, instead of mere points
of light, as in the case of the fixed stars, and may be
noticed also to pass through phases like the moon,
especially in the case of the two that are inside the earth's
orbit. The variability of their brightness is caused partly
by this, partly by change in their distances from the earth.
The well-known rings of Saturn are now supposed to con-
sist of a shower of meteorites revolving round him.

Supposing us to be situated in the northern hemisphere,

and not too far north, if we watch the apparent motions of
the heavenly bodies in the sky we shall notice the following
facts. The sun rises latest and sets earliest about the
21st of December, while the opposite is the case about
the 21st of June. During the winter half of the year his
risingand setting is south of the east and west points of
the horizon, and during the summer half they are north
of it while at two intermediate periods, known as the
;

equinoxes, he rises due east, remains in sight for 12 hours,

and sets due west. At midwinter the arc he describes
through the sky is the lowest, and at midsummer the
highest.
When the moon is first seen as a young moon she is a
little to the east of the sun. She rapidly moves through
 The Moon.

the sky towards the east, so that about full moon she
rises as the sun sets, and later on is seen as a crescent

rising before the sun in the early morning. The height

t
^ " rnv h ih^ ri^r in th Q
r
will bf rhnftrvprl t^ b^ (nnlilrr-
nky
th e case of the oun) quite indopondont of the time of
year . The interval between two new moons that i3 th e
time she-ta-kes^te-make fffi-apparent-ek cuit of th e-sky is
about 2#fdays and she rises each day about three quar --^*^-//*
;

^ a^ CUl
t er s of an hour later than the day before.
^y
The observed, will be noticed to rise
stars, if carefully
each night a than four minutes earlier than they
little less

did the night before, so that at any given hour a certain

portion of the sky which was visible at the same hour the
night before will have disappeared in the west, and a
similar portion will have come into view in the east. In
fact the whole mass of the stars appears to be slowly over-

taking the sun (or rather the sun to be moving through

the stars) and, as a consequence,
; if the stars
were visible in the day time this motion could be
plainly seen. The points of rising and setting of the stars
are always the same. The sun and all the stars reach
their greatest height in the sky or culminate, as it is
termed at a point where they are due north or south of
the spectator.

Thestars in the northern portion of the sky, from the

horizon up to a certain point depending on the position
of the observer, never rise or set, but describe in the
twenty-four hours concentric circles round an imaginary
point called the pole, and in a direction contrary to that
of the hands of a watch.

The planets, if carefully observed, will be

different

noticed, not only to change their positions among the

fixed stars, but to vary in brightness from time to time.

So much for the apparent motions of the heavenly

bodies. We have now to consider their real ones.
 The Earth's Motion,

The earth describes an elliptic orbit round the sun in

about 365 \ days. It also revolves on its own axis in about
a day. This axis remains parallel to itself and is in-
clined to the plane of the orbit at an angle of about 23
27'. Hence the phenomena of the seasons, and of the
varying positions of the sun from day to day.

Figure 1 shows the position of the earth with reference

to the sun at the different seasons. N is the north pole,
S the south and A a point in the northern hemis-
pole,
phere. The hand sphere shows the earth's position
left

when it is midwinter at A, and the right hand sphere

when it is midsummer.
The motion of the earth round the sun causes the latter
to continually change its apparent position amongst the
stars. Its path through them is called the ecliptic, and
lies,of course, in the plane of the earth's orbit. The
earth's revolution round its own axis, although on an

average 24 hours if taken with reference to the sun, really

takes place in space in about 3 minutes and 56 seconds

less than 24 hours, the difference being, in fact, the same
as that between two successive risings or settings of the
same star. It should also be noted that, owing to the
enormous distances of the fixed stars from us, all lines
drawn from the earth, no matter what its position, to
any star, are sensibly parallel.
 Mean time.

Fig. 2.

In figure 2 S the sun and E,, E 3 the position of the

is

earth on two successive days when it is apparent noon at

the point A. 1
and 2 are the earth's centre, and S t
S 2 the apparent positions of a certain star as seen from
A, Oj S t being parallel to O y S. Let S 2 2 intersect
the earth's circumference at B. It is evident that when

the earth's revolution ^oh its

Ims^has^broughTA to the

position B in E 2 that it will have described a complete

revolution with reference to space and that the star will
again be on the meridian in other words that a sidereal

day willhave elapsed since it left the position E and ;

t

that to bring the sun on to the meridian at A in E 2 it

will have to describe an additional arc B A. This arc is
the same as the angle B O s A, which is equal to the angle
2 S Oj, and is the difference between a solar and a
sidereal day. If there are n days in the year the value of

this arc will be . Reduced to time it is about 3m. =;6s.

Ordinary clocks are arranged to keep such a rate that

24 of their hours give the average interval between two
successive culminations of the sun. The real interval is,

however, sometimes more, sometimes less than this : and,

consequently, the sun does not culminate or pass the
meridian at noon, but sometimes before it, sometimes
after it, the greatest difference being about l6j minutes.
The instant the sun is on the meridian is called "apparent
 The equation of time.

noon." Noon as shown by a perfect clock is called

"mean noon." The interval between the two is called the
"equation of time." Its greatest amount is about the ist
of November, when the sun culminates about nh. 43m.
41s. A.M. The equation then diminishes till about the 24th
December, when mean and apparent noon coincide. After
that the equation increases (the sun culminating after
noon) till it attains a maximum of 14J minutes about the
nth February, and then continues to decrease, becoming
zero again about the 14th of April. It attains a maximum
of 3m. 50s. about 14th May, becomes zero about 14th
June, 6J minutes about 25th July^and zero 31st August.
The cause of the equation of time is as follows. If the
earth moved round the sun in a circle and at a uniform
rate, and if the axis on which it itself turns were perpen-

dicular to the plane of its orbit, the sun would culminate

each day at noon exactly. But the earth moves in an
ellipse and at a variable fate, and its axis is inclined to
the plane of the ecliptic at a considerable angle, the com-
bined effect being that we have the equation of time.

The great circle on the earth whose plane passes

through the centre and is at right angles to the axis is
called the "equator," and the projection of its plane in the
heavens is also called the equator, and sometimes the
equinoctial. If the sun, in its apparent annual path,
moved at a uniform rate and traversed the equinoctial in-
stead of the ecliptic we should have no equation of time.
An imaginary sun moving in this way is called the "mean
sun."
In addition to the time kept by an ordinary clock and
that kept by the sun in other words "mean time" and

"apparent solar time" we have a third kind called

"sidereal time," that is, the time kept by the stars. It

has been already mentioned that the interval between two

successive culminations of the same star is a little less
than 24 hours the time it takes, in fact, for the earth
;
io Sidereal tunc.

to make a single revolution on its axis. If we divide this

interval into 24 equal parts we have 24 sidereal hours; and
if we construct a clock with its hours numbered up to 24
instead of 12, and rate it to keep time with the stars, it is
easy to see that the hour it shows at any instant will give
the exact position of the stars in their apparent diurnal
revolution round the earth. Clocks and chronometers of
this description are used the former in fixed observatories ,

the latter for surveying purposes.

The
subject of sidereal time will be referred to later on.
Before proceeding further it will be necessary to explain
the meaning of the various astronomical terms in ordinary
use.
 CHAPTER II.

EXPLANATION OF CERTAIN ASTRONOMICAL TERMS. THE

NAUTICAL ALMANAC.

For practical purposes the earth maybe considered as a

stationary globe situated at the centre of a vast transparent
sphere at an infinite distance to which are attached the
fixed stars, and which revolves round it in a little less than
24 hours. The sun, moon, and planets appear to move on
the surface of this great sphere, the sun in the ecliptic,
the rest in their respective orbits.

The extremities of the earth's axis are called the poles ;

and the poles of the great sphere are the points where the
axis produced meets it.

Great circles passing through the poles are called

"meridians." This term applies both to the earth and
the great sphere. In the case of the latter they are also
called "declination circles." Meridians are also called
"hour circles," and the angle contained between the
planes of any two meridians is called an "hour angle," be-
cause it is a measure of the time the sphere takes to
revolve through that angle. It follows that the hour

angle is the angle formed by two meridians at the poles.

In speaking of the meridian of a place we mean the
great circle passing through the place and the poles; and
a great circle passing through the poles of the great
sphere and the zenith (or point in the sky immediately
12 Latitude and longitude.

over the observers head) is the meridian for the instant,

as regards the great sphere.

To fix the relative position of points on the earth's sur-

face we employ certain co-ordinates, called "latitude" and
"longitude." The former is the angular distance of any
point from the equator, and is measured along a meridian
north or south as the case may be. The latitude thus
varies from zero at the equator to 90 at the poles.

Longitude is the angular distance of the meridian of the

place from a certain fixed initial meridian, and is measured
either by the intercepted arc of the equator or by the
angle contained by the two meridians. Longitude is
measured east for 180 and west for 180 Different .

countries reckon from different initial meridians. The

English use that of Greenwich. The present system has
many inconveniences, and it is to be hoped that some day
the world will unite in adopting some fixed meridian and
will reckon longitude through the whole 360 degrees in-
stead of as at present.

The
position of the heavenly bodies on the great sphere
isdetermined by similar co-ordinates, but the latter arc
called "declination" and "right ascension," the former

corresponding to latitude and the latter to longitude.

Declination is measured from the equinoctial towards
the poles, and right ascension eastward from a certain
meridian. The latter is, however, reckoned through the
c
whole 36o ',and is counted by hours, minutes, and seconds
instead of by degrees, 1 hour corresponding to 15 degrees.

The point where the zero or 24-hour meridian cuts the

equator is called the "first point of Aries," and is desig-
nated by the symbol Y. It is also one of the intersections
of the equator with the ecliptic. On referring to the Nautical
Almanac it will be seen that the co-ordinates of the stars
are continually changing. The fact is that, owing to the
slow conical motion of the earth's axis known as the "pre-
 Altitude and azimuth. 13

cession of the equinoxes," the planes of reference are

changing. This, however, causes no practical inconveni-
ence, as the relative positions of the stars remain the same.

It should be noticed here that the terms "latitude" and

"longitude" are also used with reference to the heavenly
bodies, and are liable to cause confusion. These co-
ordinates are measured from and along the ecliptic, and
are not required for the problems here treated of.

Besides the above-mentioned co-ordinates which relate

to the relative position of points on a sphere another set
is necessary to fix the position of a heavenly body with

reference to the observer at any instant. They are called

"altitude" and "azimuth." The first scarcely needs ex-
planation. The second the angle formed by the verti-
is

cal plane passing through the observer and the object

with the plane of the observer's meridian. The altitude
and azimuth of a star at any instant are, in fact, the angles
read by the vertical and horizontal arcs of a theodolite
respectively when the latter has been clamped with its
zero due north, and the telescope has been directed on the
star. Azimuth is generally reckoned from the north
round by the east, south, and west ; but it is sometimes
reckoned from the south.

The plane of the "sensible horizon" is the horizontal

plane passing through the observer's position, and there-
fore tangential to the earth's surface at that point. The
"rational horizon" is a plane parallel to that of the sen-

sible horizon and passing through the centre of the earth.

The projections of these two planes on the great sphere
coincide, being at an infinite distance.

It easy to see that about half the great sphere is in

is

sight at any instant. The portion that is visible depends

generally on the latitude of the place and the sidereal
time of the instant. At the north pole the whole north-
ern hemisphere would be always in sight and no other
14 Definitions.

part. At the south pole the view would be limited to the

southern hemisphere. At the equator both poles would
be on the horizon, and every point on the great sphere
would come in sight in succession. At intermediate
places a certain portion round one pole would always be
above the horizon, while another portion round the other
pole would never be visible.
"Parallels of latitude" are small circles made by the in-
tersection with the earth's surface of planes parallel to the

equator. Similar circles on the great sphere are called

"declination parallels." A little consideration will show
that within a certain distance of the equator at each side
of it the sun will, twice in the year, pass overhead
at mid-day. The belt enclosed between the two parallels
within which this takes place is known as the "tropics."

A few more technical terms require explanation. When

speaking of the "hour angle" of a heavenly bodv at any
instant we mean the angle formed at the pole by the
meridian circle of the instant and the declination circle
passing through the body.

By the term "circumpolar star" is meant a star which

never sets but appears to describe a complete circle round
the pole. These stars cross the merdian twice in the
twenty-four hours. One crossing is called the "upper
transit," the other the "lower transit." At the points be-
tween the transits at which the stars have the greatest
azimuth from the meridian they are said to be at their
"greatest elongation," either east or west.
The words "transit" and "culminate" have the same
meaning when used with reference to stars which rise and
set.

"Parallax" is the change in the apparent relative posi-

tion of objects owing to a change in the observer's posi-
tion. Astronomically it generally signifies the difference
in the apparent position of a heavenly body as seen by an
 Parallax. 1 5

observer from what it would be if viewed from the centre

of the earth. Parallax is greatest when the object is on
the horizon, and nothing when it is in the zenith. The
moon, from being near the earth, has a considerable
parallax. That of the sun does not exceed 9". The
positions of the sun, moon, and planets given in the Nauti-
cal Almanac are those which they would have as seen
from the earth's centre, and it is therefore necessary to
correct all observations on those bodies for parallax.

Parallax causes the object to have less than its true

altitude. Refraction has the opposite effect. The latter,
like the former, diminishes with the altitude. Near the
horizon say within 10 degrees of it its effect is very un-
certain, and observations of objects in that position are
therefore unreliable. At an altitude of 45 the refraction
is about 1'. As it varies with the temperature and atmos-
pheric pressure the barometer and the thermometer must
be read if very exact results are required.

The corrections for refraction and parallax are not to be

found in the Nautical Almanac, but are given in all sets of
mathematical tables. The N. A., as a rule, gives only
variable quantities such as declination, right ascen-
sion, equation of time, etc. It is rather a bulky volume, but
the portions of it in general use by the practical surveyor
could be comprised in a small pamphlet. The most use-
ful are the sun's declination and right ascension, the equa-

tion of time, the sun's semi-diameter, and the sidereal

time of mean noon in the year
all given for every day ;

the declinations and right ascensions of the principal

fixed stars, taken in regular order according to their right
ascensions and the tables for converting intervals of mean
;

time into sidereal time and vice versa. To these may be

added tables of moon-culminating stars, and tables for

finding the latitude from the altitude

of the pole star when
off the meridian.
16 The Nautical Almanac.

Specimens of the two first pages of the quantities

given for each month in the Nautical Almanac, and of the
data for fixed stars, are reprinted below.All the quanti-
ties are given for noon at Greenwich on the day in ques-
tion. They must, therefore, be corrected by a propor-
tion for any other hour or longitude. Thus, when it is
noon go west longitude, or six hours west of
at a place in

Greenwich, 6it
p.m. at the latter. Therefore, if an
is

observation were taken at the western station at noon the

quantities required would have to be corrected for their
change in six hours.
Owing to the earth's uniform revolution round its axis
a change in longitude affects alike mean time and sidereal
time. Thus, if the mean time at Greenwich was 3 p.m.,
and the sidereal time nh., they would be 9 a.m. and 5I1.
respectively at a place in longitude go west.
In the Nautical Almanac the day is supposed to com-
mence at noon and to last for 24 hours. Thus 9 a.m. on
the 2nd of January, ordinary civil reckoning, is 2ih. of the
1st of January. This astronomical method of reckoning
mean time must not be confounded with sidereal time,
which is quite a different thing.
The data for the first of each month are given also at
the end of the preceding month. Thus, we find 32 days
given in January, the one shown as the 32d being really
the 1st of February. This is for convenience in interpo-
lating.
 Nautical Almanac. 17

JUNE, 1880.

AT APPARENT NOON.

3
i8 Nautical Almanac.

JUNE, 1880.
AT MEAN NOON.

i
 The Celestial Globe. 19

To revert to the subject of sidereal time: Since the

sidereal clock stands at zero or 24I1. at the instant the 1st

point of Aries is on the meridian, and as the clock keeps

time with the stars in thi& apparent diurnal revolution round
the earth, it follows that when any particular star is on the
meridian its right ascension is the sidereal time of the in-
stant. Thus, if the stars R. A. were 6h. the clock should
show that time at the instant of the stars transit, and its
error may be ascertained by mounting a telescope so as to
move only in the plane of the meridian, and noting the
instant of transit. If we want to find the mean time of a
star's transit we have only to convert the star's R. A.
into the corresponding mean time of the instant, in the
manner to be presently explained. Conversely, a star's
transit gives us the sidereal time of the instant, and hence
the true mean time.

The celestial globe is of great use in studying astronomy.

It is model of the great sphere supposed to be viewed
a
from outside. The positions of the stars on it are the
points where straight lines, joining them with the earth,
would intersect it. The equator and ecliptic the latter
being the sun's annual path through the stars are marked
on it, as also the sun's place in the ecliptic for every five
days. The axis on which it turns is that of the poles.
The metal ring passing through the latter represents the
meridian, and the flat horizontal ring the plane of the
rational horizon.
One of the chief uses of the globe is to show the posi-
tion of the stars at any instant with regard to the spec-
tator. To do this we raise the pole by means of the
graduation on the meridian so as to give it an altitude
above the horizon equal to the latitude of the place, and
bring the sun's place in the ecliptic for the day to the
meridian. The half of the globe above the horizon will
now roughly represent the position of the visible hemis-
phere at noon. To find the position of the sphere at any
20 The Great Sphere.

other hour it is only necessary to turn the globe through

an hour angle equivalent to the number of hours from

noon. Thus, if we wanted to find out the visible positions
of the stars at 8 p. m. we should have to revolve the globe
westwards through an angle of 120. Conversely, we can
find the name of any constellation or star by noting
its position in the sky and the hour, and setting the globe
accordingly.
The great circle passing through the zenith and the
east and west points of the horizon, and therefore at right

angles to the meridian, is called the ''prime vertical."

Pig- 3-

In Figure 3 the small circle at the centre represents the

earth, and the large circle the great sphere. Strictly

speaking the former should be a mere point in comparison

to the latter, and the points on the great sphere would ap-
 Explanation of Terms. 21

pear at A, the spectators position, in the same places as if

viewed from O, the earth's centre, h r is a tangent to the
earth's surface at A, and therefore lies in the plane of the
observer's horizon. HBR the plane of the rational
is

horizon, and p p the

1
earth's polar axis, meeting the
great sphere in the points P P1 . Z is the zenith. N
the nadir, or point on the sphere diametrically opposite
it. The plane of the paper represents the plane of the
observers meridian, and H, R, are the north and south
points of the horizon, e q is the equator, E y Q the
equinoctial, y the first point of Aries, and P y P
1
the
initial declination circle passing through it, from which
all right ascensions are reckoned. S is a star situated on
it, and S 1 another star on the observer's meridian. Z S B
isa portion of a great circle, passing through the zenith
and S and meeting the horizon at B. Z B is of course
90 The arrows represent the apparent motion of the
.

great sphere with respect to the earth.

The arc Acr or the angle A O q, is the latitude of the
,

point A, and A O q Z O Q, which is the zenith distance

of the point where the equator cuts the meridian. Also,
Z O Q=r9o P OZ=POH;
or the altitude of the visi-
ble pole above the horizon is the latitude of the place.
It should be noticed that the whole of the hemisphere
above the plane H B R is visible to the observer at A.
The S is zero, its declination
right ascension of the star
(north) S y, hour angle S P Z, its altitude S B, zenith
distance S Z and azimuth S Z R. The star S 1 has R. A.

y Q, declination (south) Q S', hour angle nil, altitude

S R, zenith distance, S Z, and azimuth zero.
1 1
The
sidereal time of the instant is y P Q, or the arc y Q. It

is therefore the same as the R. A. of the meridian. The

triangle P Z S is called the "astronomical triangle." It

should be noted that in all calculations if north declina-

tion is reckoned positive, south declination must be counted

negative, and vice versa.
 CHAPTER III.

Uses of practical astronomy to the surveyor.

Instruments employed in the field. Methods
of using them. taking altitudes. problems
relating to tim]{.

The principal uses of practical astronomy to the sur-

veyor are that it enables him to ascertain his latitude,
longitude, local mean time, and the azimuth of any
given line the latter of course giving him the true north and
;

south line and the variation of the compass. In fact the

only check he has on his work as regards direction when
running a long straight line across country is by determin-
ing its true azimuth from time to time, allowing (as will
be explained hereafter) for the convergence of meridians.
The instruments usually employed are the transit theo-
dolite, sextant or reflecting circle with artificial horizon,
solar compass, portable transit telescope, and zenith tele-

scope. To these must be added a watch or chronometer

keeping mean time, a sidereal time chronometer (this is
not, however, absolutely essential), the Nautical Almanac
for the year, and a set of mathematical tables. With the
sextant or reflecting circlewe can measure altitudes and
work out allproblems depending on them alone, and also
lunar distances. The transit theodolite may be used for
altitudes, and also gives azimuths. The solar compass
is a contrivance for finding, mechanically, the latitude,
 Instruments. 23

meridian line, and sun's .hour angle. The zenith tele-

scope gives the latitude with great exactness, and is par-
ticularly suited to the work of laying down a parallel of
latitude. The transit telescope enables us to determine
the mean and sidereal time, latitude, and longitude. The
transit theodoliteanswers the same purpose, but is not
so delicate an instrument. It is, however, of almost uni-
versal application,and nearly every problem of practical
fieldastronomy may be worked out by its means alone if
the observer has a fairly good ordinary watch. The sex-
tant has been called a portable observatory but in the ;

writer's opinion the term is more applicable to the last

named instrument. The sextant is not so easy to manage
and only measures angles up to about 116, so that 58 is

practically the greatest attitude that can be taken with it

when the artificial horizon has to be used. The latter,
as generaliy made, is disturbed by the least wind, and
then gives a blurred reflection, maWng the observation
nearly worthless. There is little use in having the arc
graduated to read to within a few seconds if the contact
of the images cannot be made with certainty to within a
minute or two.
All observations taken with the transit theodolite should,
if the nature of the case admits of
it, be repeated in re-

versed positions of the telescope and horizontal plate, and

the mean of the readings taken, as we thereby get rid of
the effects of collimation, index, and other instrumental
errors. Thus, for an altitude, the plate having been
levelled, the vertical arc set at zero, and the bubble of the
telescope level brought to the middle by the twin screws,
the verticality of the axis is tested by turning the upper
plate in azimuth 180, and seeing if the bubble is still in
the centre. If it is not it is corrected, half by the lower

plate screws, half by the twin screws, and the operation

repeated till the bubble remains in the centre in every
position. The altitude is then taken, the telescope
24 A Ititudes.

turned over, the upper plate turned round, and the alti-
tude again read. In each case both verniers should be
read.
The step after taking an altitude with either
first

sextant or theodolite is to correct it for index error, if

there is any. The following lists give the corrections to
be applied in each case to an altitude of the sun's upper
or lower limb to obtain that of his centre :

THEODOLITE. SEXTANT.

Altitude above Double Altitude with

water horizon. artificial horizon.

Index error. Index Error. Index Error.

Dip of Horizon. Divison by 2.
Refraction. Refraction. Refraction.
Parallax. Parallax. Parallax.
Semi-diameter. Semi-diameter. Semi-diameter.
The semi-diameter has to be added if the lower limb
isobserved, and vice versa. When taking an altitude for
time with the artificial horizon the easiest way to get the
correct instant of contact is to bring the two images into
such a position that they overlap a little while receding
from each other. At the instant they just touch the
observer calls "stop," the assistant notes the exact watch
time, and the vernier is then read. This plan necessitates
observing the lower limb in the forenoon and the upper
in the afternoon. The dip depends on the height of the
instrument above the water, and, like the refraction and
parallax, is to be found in the mathematical tables.

In the case of a meridian altitude for latitude the sun

or star, after rising to its greatest height, appears for a
short time to move horizontally. When this is the case
the altitude may be read off.
Fixed stars require, of course, no correction for parallax
or semi-diameter. As the refraction tables require a
correction for temperature and atmospheric pressure the
height of the thermometer and barometer should be noted.
 Equation of Time. 25

If an altitude has to be taken with the sextant and

artificial horizon, and the sun is too high in the heavens
for the instrument, a suitable star must be observed in-
stead.

In surveying operations the latitude is generally known

approximately. This gives the approximate altitude for
a meridian observation; for the altitude of the intersec-
tion of the meridian and equator being 90 minus the
latitude, we have
only to add to or subtract from this
quantity the objects declination, and we have the al-
titude.
ZPt'y ftct/r-rs/- CAUSE OF THE EQUATION OF TIME.

In Figure 4 P is the pole, E C

a portion of the ecliptic, and E Q
a portion of the equator each;

being equal to go. C and Q are

on the same meridian, and P Q is
also a quadrant. Now, let S be the
sun, and suppose it to move at a
uniform rate from E to C. Let Fig. 4.
S 1 be an imaginary sun (called the ''mean" sun) moving
in the equator at the same rate as the real sun. Now, let
the two suns start together from E, and after a certain
interval let their position be as shown in the figure.
Since they move at the same rate, E S will be equal to
E S 1 but as a consequence the meridians P S and P S
,
1

will not coincide, S 1 having got ahead of S. The angle

SPS 1 formed by the two meridians is the equation of
time. As the two suns must arrive simultaneously at C
and Q it is evident that, though S 1 gains on S at first, it
will, after a certain point, cease to gain and lose instead. %

Since the equation of time in other words the differ-

ence between apparent and mean solar time is con-
tinually changing, if we want to find from the Almanac
the mean time corresponding to apparent time at any
26 Connexion of Sidereal

particular instant and longitude, we must allow for the

change in the equation that has taken place since noon
at Greenwich. For instance ; suppose we had to find the
mean time corresponding to three hours p.m. apparent
time on the 22nd April, 1882 at a place in longitude 6h.
west. By the N. A. the equation of time at apparent noon
that day at Greenwich was im. 34s. 43, to be subtracted
from apparent time and increasing, the variation per hour
o.s.496. At 3 p.m. at the place it would be 9 p.m. at
Greenwich. 9 x 05.496=45.464. The corrected equation
of time is im. 38s. 89, and the true mean time 2h. 58m.
2IS.II P.M.
GIVEN THE SIDEREAL TIME AT A CERTAIN INSTANT TO
FIND THE MEAN TIME.
Here we have given the right ascension of the declina-
tion circle of the great sphere that is on the meridian at
the instant, or which is the same thing the time that a
sidereal clock would show. Now the Nautical Almanac
gives the sidereal time of mean noon at Greenwich, which
has to be corrected for longitude. These two data give
us the interval in sidereal time that has elapsed since
mean noon, and this, converted into mean time units,
will be the mean time.
Ex. Find the mean time corresponding to 14 hours
sidereal time at Kingston on the 28th April, 1882.
We find from the N. A.
Sidereal time of mean noon at Greenwich 2h. 25m. 25S33
Correction for longitude 50 26

Sidereal time of mean noon at Kingston 2h. 26m. 15s 59

Sidereal time of the instant 14I1. om. os

Difference, or interval of sidereal time that has elapsed

since mean noon nh. 33m. 44S41
Which, converted into mean time, is 1 ih. 31m. 50S75 p. m.

The conversion of sidereal into mean time units, and

vice versa, is obtained from tables at the end of the Nauti-
cal Almanac.
 and Mean Time. 27

If the sidereal time of mean noon is greater than the

siderealtime given we shall obtain the interval before
mean noon. Thus, if on the same date as above we
wanted to find the mean time corresponding to sidereal
time one hour we should proceed as follows :

Sidereal time of mean noon

" of the instant 100
2h. 26m. 15S69

Sidereal interval before mean noon 1 26 15- 69

Which in mean time units is 1 26 1 56

Subtracting this from I2h 12 o o

We have mean time ioh. 33m. 585-44 a.m.

sometimes convenient to add 24 hours to the given

It is

sidereal time to make the subtraction possible. Thus, if

the sidereal time were ih., and the sidereal time of mean
noon 23I1., we should have the interval elapsed since mean
noon 25 23, or 2 hours, which is ih. 59m. 40S.3 p.m.,
mean time.
TO FIND THE MEAN TIME AT WHICH A GIVEN STAR WILL
BE ON THE MERIDIAN.
This is only an application of the preceding problem.
For the Almanac gives us the star's right ascension, which
is the same thing as the sidereal time of its culmination,

and we have merely to find the mean time corresponding

to it.

GIVEN THE LOCAL MEAN TIME AT ANY INSTANT TO FIND

THE SIDEREAL TIME.
Here we must convert the interval in mean time that
has elapsed since the preceding noon into sidereal units,
and add to it the sidereal time of mean noon.
Ex. Find the sidereal time at 9 a.m., pn the 29th of
April, 1882, at Kingston, Canada.
Here we have, as before :

Sidereal time of mean noon on the 28th 2I1. 26m. 155-69

Add 21 hours of mean time in sidereal units, or 21 3 26-98

Sidereal time 23b. 29m. 425-67

2$ Sidereal Time.

If thisprocess makes the result more than 24 hours

that number must,of course, be subtracted from it. Thus,
if we got 25 hours the sidereal clock would show ih. If

the sidereal time of mean noon is greater than the inter-

val in sidereal units we add 24 hours to the latter to make
the subtraction possible.

Thecorrection on account of longitude for the sidereal

time of mean noon is constant for any particular place or
meridian
The subject of sidereal time may be thus illustrated :

In Fig. 5 let the small circle

represent the earth, and the
large circle the equator of the
great sphere viewed from the
north, the plane of the paper
being the plane of the equator.
Let P be the pole, A a point on
the earth's surface, and P A
the meridian of A. y is the
1
point of Aries, S and S
first Fig. 5.
two stars, and s and s the points where their declina-
l

tion circles meet the equator. Now the arc y s1 (or the
angle y P)
s1 is the right ascension of S lf and the arc
y s 1 s that of S. Now suppose the earth (and therefore
the meridian P A m)
to remain fixed, while the outer
circleand stars revolve around it in the direction of the
arrow; and at the instant that it is mean noon on a cer-
tain day at A let the position of the great sphere be as
shown in the figure. The arc y s l m will be the sidereal
time of mean noon for that day at A. The star S will be
on the meridian at an interval of sidereal time after mean
noon corresponding to s m, while the star S 1 has passed
the meridian by an interval corresponding to m $ tt and
by reducing these intervals to their equivalents in mean
time we shall have the mean times of their transits. For
 Hour Angle of a Star. 29

instance, suppose we had to find at what time the pole

star would be at its upper transit on a day when the
sidereal time of mean noon
was 2ih. 30m., the right as-
cension of the star being taken
as ih. 15m. Now the state
of things at noon would be as
shown in Fig. 6. The star
would have passed the meri-
dian by an interval of 2ih.
30m. ih. 15m., or 2oh. 15m.
(sidereal) and would there- Fig. 6.

fore be on the meridian at 3I1. 45m. sidereal, or 3h. 44m.

23s. mean time after noon.

Sidereal time usually found by calculating the hour

is

angle of a star from its observed altitude. This, added to

the star's right ascension if the hour angle is west, or
subtracted from it if east, gives the sidereal time. From
this the mean time can be obtained, if required. The
watch time at which the altitude is observed must, of
course, be noted.
TO FIND THE HOUR ANGLE OF A GIVEN STAR AT A GIVEN
TIME AT A GIVEN MERIDIAN.
Here we must find the local sidereal time of the given
instant and take the star's right ascension from the Al-
manac. The difference between these two quantities
will be the star's hour angle, which will be east if the
star's R. A. is greater than the sidereal time, and west if
the contrary is the case.
TO FIND THE MEAN TIME BY EQUAL ALTITUDES OF A
FIXED STAR.
Fixed stars are employed for this purpose in preference
to the sun or planets because of the change in declina-
tion of the latter. A star should be chosen which de-
scribes a sufficiently high arc in the sky. Two or three
hours before its culmination its altitude is taken with the
30 Mean Time by cqnal Altitudes.

sextant or theodolite, the exact watch time noted, and

the instrument left clamped at that altitude. Some hours
later, when the star has nearly come down to the same
altitude, the observer looks out for it(keeping the instru-
ment still
clamped) till it enters the field of view of the
telescope, and waits till it has exactly the same altitude as
before, when he again notes the watch time. The mean
of the times of equal altitude will give the watch time of
the star's culmination, which should be the same as the
mean time (previously calculated), corresponding to the
star's right ascension, the latter being the sidereal time
of the culmination. If they are not the same the differ-
ence will be the watch error.
used for this observation the verti-
If the theodolite is
cal arc only kept clamped. When the star has nearly
is

come down^to the Gocond altitude the horizontal arc is

-

clamped and its slow motion screw used.

TO FIND THE LOCAL MEAN TIME BY AN OBSERVED ALTI-
TUDE OF A HEAVENLY BODY.
For this problem we must know the latitude of the
place, and, if the sun is the object observed, we must also
know the mean time approximately in order to correct its

declination.
The altitude should be taken when the heavenly body
israpidly rising or falling that is, as a rule, when it is
about three hours from the meridian, and the nearer to
the prime vertical the better.
If we take P as the pole, Z the zenith, and S the hea-

venly body (Fig. 7), PZS

be will
a spherical triangle in which PZ
is the complement of the latitude,

PS the polar distance of the ob-

ject observed, and ZS the comple-
ment of the altitude. The three
sides being given we can find the
three angles from the usual for-
mula*. In the present instance we Fig 7.
 Time by Altitude of Sun. 31

want P, which is the hour angle of the object.

A convenient formula is

c . P sin (s-PS) sin (s-PZ)

bin'
2 sin PS sin PZ
PZ + PS + ZS
where, s is
2
The watch time is noted by an assistant at the instant
the altitude is taken. The usual corrections are applied
to the altitude, and the angle P having been worked out
is divided by 15. This gives us the hour angle of the
body. In the case of the sun it will be the apparent solar
time, and by adding or subtracting the corrected equation
of time we shall get the true mean time of the instant of
the observation.
If the body a fixed star or planet, then, from its
is

known right ascension we subtract the hour angle if it is

east, or add it if it is west. This gives the sidereal time
of the instant, from which the mean time can be inferred
if required. The spherical triangle PZS is known as the
"astronomical triangle."
Instead of taking a single altitude two or three may be
taken in quick succession and the time of each noted.
The mean of the altitudes is then taken as a single alti-
tude to correspond to the mean of the times. If the tran-
sit theodolite is used two altitudes should be taken in re-

versed positions of the telescope and horizontal plate so

as to correct instrumental errors. Still greater accuracy

is obtained by observing both an east and a west star and

taking the mean of the results, as errors of observation,

refraction, and of the instrument, will be got rid of in a

great measure.
EXAMPLE OF WORKING OUT A SEXTANT DOUBLE ALTITUDE
OF THE SUN'S LOWER LIMB FOR LOCAL TIME.
17th April, 1882. Lat. 44 13' 40" N., Long. 5*1. 5m.
50s. W. Sun's semi-diameter, 15' 57". Declination io
40' o" Watch time of observation, jh. 37m. 15s. p.m.
N.
Equation of Time, otn. 34s., to be subtracted from appar-
ent time. Index Error, 5' 30", on
 3 Example.

Double altitude 64 4' o"

Index error 5 30

2 58 30
)63

3i 59 15
Semi-diameter 15 57

32 15 12
7'//rv>r co/- (
Refraction and parallax ... 1 23

cr/y//iv</ /*/i'J True altitude of sun's centre 32 13 49

c/ 1 *r sn ft
90

90 o o 57 46 n=ZS
Declination 10 40 o 79 20 o =PS
45 46 20 PZ=
79
 Time by Star A Ititude. 33

EXAMPLE OF CORRECTING A MEAN TIME WATCH BY A

SINGLE ALTITUDE OF A STAR.
On a certain date at a place in 5I1. 30m. west longitude
the altitude of a star was taken at 9 p.m. by the watch,
and the hour angle when worked out proved to be 2h.
30m. 17s. west.
To find the watch error
The star's right ascension was 4I1. 29m. 4s.
Add hour angle 2 30 17

Sidereal time of the instant = 6 59 21

Add 24

30 59 21
Subtract the sidereal time of
mean noon, corrected for
longitude 22 o 9

Sidereal interval since mean

noon 8 59 12
= 8h. 57m. 44s. in mean time.
And the watch was 2m. 16s. fast.
It should be noted that if the declination of a heavenly
body is north it must be subtracted from 90 to get its
polar distance. If south the declination must be added
to 90 .

The following formula is a very convenient one, as the

altitude and latitude are employed instead of their com-
plements.

Sin 2 = : : =r=r ;where a is the altitude. I the

2 cos / sin PS
latitude, PS the polar distance, and s=
TO FIND THE TIME BY A MERIDIAN TRANSIT OF A
HEAVENLY BODY.
This is a very simple method when the direction of the
true north known, as when running north and south
is

lines on a survey. The theodolite is set up on the line,

34 Time by Meridian Transits.

the telescope directed on a distant point or mark on it,

and the horizontal plate clamped. The telescope will
now, if moved
in altitude, keep in the plane of the meri-

dian, provided the instrument is in adjustment and the ;

instant of transit of any object across the vertical wire or

intersection being noted, we can deduce the true time.

As the altitude of an object at transit is equal to the alti-

tude of the intersection of the meridian and equator

plus or minus the declination of the object, we have the

equation
Altitude = 90 latitude declination.

and can set the telescope beforehand at the required

altitude. If the latter is more than 50 a diagonal eye

piece is necessary with most instruments. In the case of

the sun we may either take the mean of the observed
instants of transit of the east and west limbs, or take the
transit ofone limb and add or subtract the time required
for his semi-diameter to pass the meridian (which we
obtain from the Almanac). We
now have the watch time
of transit of the sun's centre, which takes place at appar-
ent noon, and have only to find the true mean time of ap-
parent noon by adding to or subtracting from the latter
the equation of time (corrected for longitude), when the
difference will give us the watch error.

Example At Kingston, on the 2nd of May, 1882, the

transit of the sun's west limb was observed at nh. 55m.
A. M. What was the watch error ?
Here we have
Watch time of transit of limb nh. 55m. os.
Time of the semi-diameter passing the
meridian im. 6s.

Watch time of transit of sun's centre uh. 56m. 6s.

The equation of time, corrected for
longitude, was
jm. 11.5s. to be subtracted from apparent time.
 Time by Meridian Transits. 35

Apparent time of transit of sun's centre 12I1. om. os.

Equation of time 3m. 11.5s.
True mean time of transit nh. 56m. 48.5s.
Watch time of transit 1 ih. 56m. 6.0s.

Watch slow 42.5s.

In the case of a star its right ascension is the sidereal
time of the instant of transit, and by working out the
corresponding mean time in the usual way we get the
watch error.

If a planet is to be observed we take its right ascension

from the part of the Almanac which gives its position at

Greenwich, correcting it for longitude in the
transit at way
directed in the explanations at the end.

For observing objects at night the theodolite ought to

have an illuminating apparatus to light up the wires.
The plan of throwing the light of a lantern through the
object glass is objectionable if it can be avoided.
It should be noted that the nearer the observed object
is to the zenith the less will be the effect of any error in
the direction of the north and south line, and the greater
that of one of the telescope pivots being higher than the
other.

Withsmall instruments, objects near the equator, from

moving the fastest, are preferable, while those nearest the

pole are the worst.

 CHAPTER IV.

METHODS OF FINDING THE LATITUDE, LONGITUDE, AND

MERIDIAN.
TO FIND THE LATITUDE BY THE MERIDIAN ALTITUDE OF
THE SUN OR A STAR.
The
altitude may be taken with the theodolite or sex-
tant. The approximate direction of the meridian should
be known beforehand. If the theodolite is used the in-
strument is levelled, and the telescope directed on the

object a little before it attains its greatest height. The

horizontal wire is then made to touch the object (the
lower limb if the sun is observed) and is kept in contact
with it as it rises by means of the slow-motion screw of
the vertical arc, the telescope being moved laterally as
required. When the object has attained its greatest
altitude it will remain for a short time in contact with the
wire, when the vertical arc is read off. The telescope is
then at once turned over, the upper plate reversed, and the
altitude again read. The mean of the two readings will
be the apparent altitude of the object.
When using the sextant and artificial horizon we bring
the two images into contact and keep them so by the slow
motion screw till they cease to separate, when the vernier
is read off.

The
usual corrections having been applied, and the true
meridian altitude of the sun's centre or star thus obtained,
the latitude is found as follows :
 Latitude by a Meridian Altitude. 37

1st case. Let the object observed culminate at the op-

posite side of the zenith to the visible pole.

In Fig. 8 let A be the observer's

p A
1
position, q p e a 'section
of the earth passing through A,
and the poles (p p ) and therefore
l :

in the plane of the meridian.

Let O be the earth's centre and
let e O q be perpendicular to p p
l
;

then e and q will will be the inter- F} 8

sections of the equator with the meridian. Draw A R H

touching the earth's surface at A and also in the plane of
the meridian. H R will be in the plane of the horizon
and due north and south. Let S be the object
will lie
observed and let its declination be north. Draw A P
parallel to p p and A Q parallel to e q. A P will be the
l

direction of the pole of the great sphere, and A Q that of

the intersection of the meridian and equinoctial. Join
O A and produce it to Z the zenith. A Z is at right
angles to H R, and P A Q is also a right angle. S A R is
the measured altitude of the object and S A Q its declina-
tion. Now the latitude of A is the arc A q or the angle
A O q. But A O qZ A Q, which is the declination of
the zenith. Also, Z A Q=9o-P A Z=P A H, which is

the altitude of the visible pole. Hence we have:

Latitude=ZAQ=go-QAR=90 (S AR-SAQ)
=90 altitude -{-declination.

If the objecthad south declination, as S 1 we should ,

have (since S 1 A R is its altitude and S 1 A Q its declina-

tion) Latitude=Z A Q=90-Q A R^go (S
1
+ AR
Q A S )=90 altitude
1
declination.

In working out latitudes it is always best to draw a

1
figure. In the one given p is the north and/) the south
pole.
 38 Latitude.

2nd case. If the object culminates between the zenith

and the visible pole, as at S in figure 9, its altitude will be
S A H, and we shall have :

Latitudes P A H=S A H S A P
=S A H (90 S A Q)
=altitude + declination 90 .

If the object is a star

which never sets, but de-
scribes a diurnal circle round
the pole, it will cross the
meridian twice in the 24
hours, and we may take its

altitude at what is known as

its lower transit, as at S 1 ,

Fig. 9. Here we have:

Latitude=PAH=S 1 AH
+ S A P=altitude + 90
1
Fig 9.

declination. Therefore, in the case of such a star we

have :

Latitude=star's altitude i star's polar distance the ;

positive sign being taken if the star is observed below the

pole, and vice versa.

Case when A is within the

2 can only apply to the sun

tropics. many books on astronomy the formuloe of

In
case 1 are made to apply to the sun in every situation,
whereas they manifestly fail when he culminates between
the zenith and the visible pole.

In the Nautical Almanac is given a very simple method

of finding the latitude from an altitude of the pole star
taken at any point of its diurnal circle round the pole.
The time of the observation has to be noted and the cor-

responding sidereal time calculated.

LONGITUDE.

Longitude cannot, like latitude, be measured absolutely,

 Longitude. 39

as it has no natural zero or origin, and we have to assume

an initial meridian arbitrarily, the English adopting that
of Greenwich. But the difference of longitude of two
places can always be found. The simplest method of
doing this is by comparing the local time at the two
places for the same instant. This is done by signal of
some kind or other, such as flashing the sun's rays from
station to station, or by the electric telegraph.
Since the earth revolves through 360 degrees of hour
angle in 24 hours it will pass through 15 degrees in 1 hour.

That is, when it is one o'clock in the afternoon at a certain

station it will be 2 o'clock at a station 15 degrees to the
east of it. Fifteen minutes and seconds of longitude in arc
will be equivalent to one minute and second of time respec-
tively. This applies to sidereal as well as to mean time.
That is, if the sidereal clock showed 1 hour at one place it
would show 2 hours at the other. (This apparent paradox
is worth thinking out, often a puzzle to beginners.)
for it is

Therefore, if at a preconcerted signal the observers at two

stations note the exact local time, either mean or sidereal,
the difference of the two will give the difference of longi-
tude. Ordinary watches may be used if their exact rate
and their error at any given instant are known. The
best way, however, is by telegraphing star transits. If
the eastern observer signals at the instant that a certain
star is on the meridian, and the western observer notes

the time of the signal by his sidereal chronometer and

afterwards takes the time of the star's transit at his
own station, the interval of time between the two transits,
allowing for the clock's rate, will evidently give the differ-
ence of longitude. If the eastern observer notes the time
of the transit, and the western signals the transit at his
station, the same result will be obtained, and by taking
the mean any time lost in the transmission of the signals
will be corrected ;
for it is evident that in the first case
the time lost will make the difference of longitude too
40 Longitude.

small, and in the second case it will make it too large.

By means of certain contrivances
possible to register
it is

the instant of a transit to a small fraction of a second,

and if a number of observations are taken the mean of
the results will be very near the truth.

Since an error of one second in the time will throw the

longitude out by about 360 yards in latitude 45 it is evi-
dent that for surveying purposes great care must be taken
to insure accuracy. When the local times are compared
by flashing signals a large number of observations should
be made and the results compared.

The subject of longitude will be more fully gone into

hereafter. It may be mentioned here that sailors obtain

their longitude by finding the ship's local mean time by

an altitude of the sun when the latter is about three hours
from the meridian, and comparing it with a chronometer
keeping Greenwich mean time, the latter being noted at
the instant the altitude is taken. It seems almost need-

less to remark that when using chronometers the correc-

tion for rate must always be applied.

METHODS OF FINDING THE MERIDIAN.

TO FIND THE AZIMUTH OF A HEAVENLY BODY FROM ITS
OBSERVED ALTITUDE.
This is a very similar problem to that of finding the
hour angle from an altitude; the only difference being
that, instead of finding the angle P of the triangle P Z S,
we have to find the angle Z. We have, as before, the
three sides of the triangle given, and may therefore use
the formula

S' a - Sin (sZ S) Sin (sZ P)

2 Sin Z S Sin Z P
Another formula that may be employed is :
 Azimuth by an Altitude. 41

Cos 2 Cos s Cos (s S P) Sec X Sec a

2
where a is the altitude of the object, S P its polar dis-

tance, /
,11.,
the latitude,
j
and
a +X
s=
PS !
+
2
This formula israther the most convenient of the two,
as it entails less subtraction than the other.
This problem (which one of great importance to sur-
is

veyors) r e quires, of course, an alt. -azimuth instrument,

such as a transit theodolite. The astronomical bearing
of any line Z A (Fig. 10) can be found in this way by
directing the telescope on the line or mark, taking the
horizontal plate reading, and then
turning it on the heavenly body and
taking its altitude and the horizontal
plate reading. It is better to repeat

the observation in reversed positions

of the instrument and take the mean
of the two sets of readings. The differ-
ence of the horizontal readings on the
line and the heavenly body gives the Fig IO
angle AZ
S, and the triangle P Z S when solved gives
the angle P Z S, whence we have A Z P the required
bearing, and therefore Z P the direction of the meridian.
In taking an alt-azimuth of the sun, if we take a single
altitude only we must add or subtract the semi-diameter
to get the altitude of the centre. If the telescope has a
vertical and horizontal wire the sun's image is made
tangential to both. To get the lateral correction for
semi-diameter we must multiply the Almanac value of the
latter by the secant of the altitude. Both these corec-
tions are got rid by observing the sun in opposite quad-
rants of the cross-wires. The best plan is to keep one
edge of the sun tangential to one of the wires by means
of a slow-motion screw till the other edge becomes also
tangential. Thus, we might let the sun overlap the
vertical wire a little and keep it tangential to the
42 Azimuth by an Altitude.

horizontal wire by the vertical slow-motion screw till it

just touches the former. If

the wires of the theodolite are
arranged as in Fig. n
we take
the observation as follows.
Suppose the time to be fore-
noon and the apparent motion
of the sun in the direction of
the arrow. For the first ob-
servation get the sun tangen-
tial to the wires a b and e f

m the uppper position. This Fig. "

is done by making it overlap a b and using
a little,
the vertical arc slow-motion screw to keep the lower
edge tangential to the horizontal wire e f until the sun also
touches a b, when the verniers are read off. The instru-
ment is then reversed and the sun made tangential to the
two wires in the lower position, this time using the
horizontal plate slow-motion screw to keep the edge tan-
gential to a b. The mean of the two altitudes is taken
and also that of the two horizontal readings. The time
must also be noted so as to correct the declination. The
reading of the horizontal plate when the telescope is
turned on the referring mark (A) is taken both before and
after the sun observations.

Ex. At Kingston in latitude 44 13' 40" on the 3rd of

March, 1882, at 2h. 30m. p.m., or 7h. 36m. Greenwich
mean time, two altitudes were taken of the sun with a
transit theodolite in reversed positions for the purpose of
testing the accuracy of a north and south line, the hori-
zontal arc being first clamped at zero, and the telescope
directed northwards along the line.
READINGS ON SUN.
Altitude. Azimuth.
1st observation 31 8' 220 o'
" 220 16
tnd 30 16
Mean 30 42 220 8
 Azimuth by an Altitude. 43

To Correct the Altitude. To Correct the Declination.

30 42' o" Declination at _
Meanj^o 44 ^7 b ,

Refraction 1 36 Noon at
Greenwich]
Parallax-] 8 Correction for 7 J hrs.j
31 ^
57". 5 per hour
'
Truealtitude30 40 32 at J

True declination 6 37 6 S
90
Sun's N. P. D.=g6 zi 6
Formula used :

Cos. 2 =cos. 5. cos. (s S P) sec. X sec. a

2
a= 30 40' 32" log cos s= 8.8687314
X= 44 13 40 log cos (s SP)= 9.9921540
S P= 96 ^y 6 log sec a =10.0654637
log sec X =10.1447380
2)171 31
2)39.0710881

I9-5355440
=10 + log cos 69 56'
2
44 Meridian by Equal A Ititudes of a Star.

= cos 180-PZ-ZS-hPS
2

go PZ + go-ZS
- + PS
cos '

2
^4-a+PS
=cos =cos s
2

tdcn oo-PZ + oo-ZS-PS

cos t
(s PS)= cos 2
= cos ( PZ + PS-fZS) =
190 ^
sins

PU
r r
Inerefore
o
cos 2
PZS =
2
cos s
~
cos
cos / cos a
(s PS)-

Similarly it
may be shown that
.
9 ZPS == cos 5 sin (sa)
sin 2
'
.--
2 cos / sin PS
TO FIND THE MERIDIAN BY EQUAL ALTITUDES OF A STAR.
Select a star which describes a good large arc in the
sky, and having levelled the theodolite direct the tele-
scope on it about two hours before it attains its greatest
height. Clamp both arcs, and by means of the slow mo-
tion screws get the star
exactly at the intersection of the
wires. Having taken the reading of the horizontal arc,
leave the vertical one clamped, loosen the upper horizon-
tal plate, and look out for the star when it has nearly
come down again to the same altitude. ^
When it enters
the field with the telescope, using the
of view follow it

horizontal slow-motion screw, but still keeping the verti-

cal arc clamped, till it is exactly at the intersection of the
wires. Now read the horizontal plate the mean of the :

two readings will give the direction of the meridian. Set

the plate at that reading and send out an assistant with a
lantern. Get the latter exactly at the intersection of the
wires, and drive in pickets at the lantern and theodolite
station. .

This method is rather a tedious one, but,, it may be

shortened by observing the star when nearer the meridian.
 Meridian by Transit of Pole Star. 45

TO FIND THE MERIDIAN BY AN OBSERVATION OF THE POLE

STAR AT ITS MERIDIAN TRANSIT.
Ascertain the watch error by any of the ordinary
methods and calculate the exact instant at which the pole
star will be on the meridian, either above or below the

pole. If the theodolite telescope is directed on it at this

instant we shall evidently have the meridian line, provid-
ed the instrument is in good adjustment. But it is better,
in order to eliminate instrumental errors, to proceed as
follows at some definite time before the star will be on
:

the meridian say 2 minutes direct the telescope on it

and take the reading of the horizontal plate. Reverse
the telescope and horizontal plate, and direct the former
on the star at the same interval of time after the transit
in this instance 4 minutes after the first observation and
again read the plate. The mean of the two readings will
give the true north.

It is almost unnecessary to remark that in observations

of this kind both verniers should be read.

TO FIND THE MERIDIAN BY THE GREATEST ELONGATION

OF A CIRCUMPOLAR STAR.
This is a very accurate method. Stars which, like the
pole star, are very near the pole, owing to their slow
motion appear to move vertically for some little time
when at their greatest eastern or western elongation. We
will suppose the pole star to be the one observed. About
six hours before or after the transit (the time of which
must be previously calculated) the theodolite is carefully
levelled, its telescope directed on the star, and the hori-
zontal plate The operation is then repeated in
read.
reversed positions of the instrument and the mean of the
two readings taken. If we have previously taken the
horizontal plate reading when the telescope was turned
on some well-defined distant object as a referring work
we can now obtain the azimuth of the latter as follows :
46 Meridian by Greatest Elongation.

In Fig. 12 let the plane of the paper

represent the plane of the horizon, and
let Z be the observer's position, A the

referring mark, P the pole, and S the star

at its greatest eastern elongation. P Z S
will be a spherical triangle, right-angled
at S, and we shall have :

_. Sin P S
Sin P Z S=xi
Sin ^-^
P Z
or, if 8 is the star's declination and A the
latitude of the place :
Fig. 12.

SinPZS=^4
COS A
since P S the complement of the declination, and
is

P Z the complement of the latitude. The latitude need

not be very accurately known.
Now, having from this equation previously found the
angle P Z S, and having obtained A Z S from the plate
readings, we get at once the angle A Z P, which is the re-
quired azimuth.
If we have to use a star some distance from the pole
we must calculate the time of its greatest elongation by
solving the equation.
Cos Z P S = cot S tan X,

which gives us the star's hour angle, and hence the time
of the observation.
The altitude is given by the equation.
sin X
Sin. altitude:
sin d
If it inconvenient to observe the pole star at its
is

greatest elongation we can use the following formula

which is approximately true in the case of a star very near
the pole.
tan A = 1
_ _
.
e
i
-
sin Z P S
tan A
where A
is the star's azimuth, Z P S its hour angle
1

at the time of observation, and A its azimuth at greatest

elongation.
 Meridian by High and Low Stars. 47

TO FIND THE MERIDIAN BY OBSERVATIONS OF HIGH AND

LOW STARS.
This a very useful method, as it is independent of the
is

pole star, and can therefore be employed in the southern

hemisphere where that star is not visible.
Choose two stars differing but little in right ascension,
one of which culminates near the zenith and the other
near the south horizon (or the north horizon if in the
southern hemisphere.) Level the theodolite very care-
fully. The great circle swept out by the collimation line
of the telescope will coincide with the meridian at the
zenith, however far it may be from it at the horizon and ;

a star culminating near the zenith will cross the centre of

the field of view at nearly the same time as if the tele-
scope had been truly adjusted in the meridian, while this
will not be the case with a star culminating near the
horizon. Having calculated the true times at which the
stars will cross the meridian observe the transit of the

upper star, noting the watch time. This will give the
watch error approximately, and we shall now know the
approximate watch time at which the lower star will
transit. By keeping the telescope turned on that star till
that arrives we shall get it very nearly in the
instant
plane of the meridian ;
and by repeating the process with
another pair of high and low stars we shall have the direc-
tion of the meridian with great exactness.

For this method we require a transit theodolite fitted

with a diagonal eye piece. The nearer the upper stars
are to the zenith the better.
The Canadian Government Manual of Survey recom-
mends for azimuth the formula :

tan P S sec X sin Z P S

~~
1 tan PStan ^cosZPS
as applied to observations of the pole star; but it re*

quires special tables in order to work it out.

48 Meridian by Pole Star.

The followingis the proof of this formula :

We have the fundamental formulae

f sin a sin C = sin c sin A (1)

_ cos c cos a cos b
I

j COS C= ; : r w
,

(2)
x

i sin a sin b
cos. a cos b cos c
[
cos A=
A
. .
sin b sin c
.
,

(3)
VJ/
v

L
From (3), cos a cos b cos 2 b cos c+sin b sin c cos b cos A
_ cos c cos a cos b
From (1 & 2), cot I = -7- -

r
x ^ csin A sin
sin a C
cos c cos 2 b cos c sin 6 cos b sin c cos A
sin b sin c sin A
_ sin b sin c sin A
.-.tan C
sin 2 b cos c sin b cos 6 sin c cos A
cosec 6 tan A c sin
1 cot b tan c cos A
In the triangle P Z S let P S=c, Z S=, and P Z=6
Z=CandP=A
Then
cosecPZtan P Ssin^PS
tan Pz S- i _ cotpZtanp g cos 2 p S ;

tan P S sec X sin Z P S

1 tan P S tan X cos Z P S
 CHAPTER V.

SUN DIALS.
To a person acquainted with the rudiments of Astron-
omy a little consideration will show that if a straight line
such as the straight edge of a thin plate of metal is
placed parallel to the polar axis of the earth, its shadow
thrown on any plane surface will, for a given hour angle
of the sun, always lie in the same straight line whatever
be the sun's declination. The shadow of any particular
point in the line will move as the declination varies, but
will always lie in the same straight line for any given hour
angle. On this principal all sun-dials are constructed.
The position of the shadow shows the sun's hour angle
at the instant, and therefore indicates the apparent solar
time ;
so that in order to obtain ordinary mean time we
have to apply the equation of time.
Dials are generally either horizontal or vertical. In
the former case the shadow of the stile, as it is called, is
thrown on a horizontal plate ;
in the latter on a vertical
wall.

The simplest forms of horizontal dials are those which

would be employed at the equator and at the poles. At the
equator the dial would evidently consist of a vertical plate
having a horizontal edge, lying north and south. The
shadow lines would be parallel to the stile, and their dis-
tances apart for equal intervals of time would rapidly
50 Sun Dials.

increase according to the sun's distance from the meridian,

and would become indefinitely great when he was on the
horizon.
At the poles the stile would be a fine vertical rod, from
the base of which 24 straight lines, radiating at intervals
of 15 degrees, would indicate the hours. The line on
which the shadow was thrown at the time corresponding
to Greenwich mean noon might be assumed as the zero
or 24-hour line. At other places the stile must be set so
that its angle of elevation above the horizontal plane is
the same as the latitude of the place.
HORIZONTAL DIALS.
A horizontal dial generally consists of a triangular
metal stile fixed on a horizontal plate on the top of a
pillar. Fig. 13 is an elevation and Fig.
14 a plan. The angle of elevation of
the stile is made equal to the latitude
of the place, and if the variation of the
compass is known, the latter may be
used to get the dial with its stile in the

plane of the meridian. The hour lines

on the plate are marked out thus let :

Fig. 13. A B (Fig 14) be the base of the stile,

Fig. 14.
 Sun Dials. 51

and A its south end. Draw A C

B A C is equal to so that
the latitude, and
any point A
at C draw C B perpen- C in
dicular to A C, meeting A B in B; produce A B to D and
make B D equal to B C. Through B draw a straight
line E B F perpendicular to A B D. From D draw lines
D a, D a D b, D b &c, meeting E B F in a, a &c,
1
,
1
,
1
,

and making the angles B D a, B D a a D b, a 1 D b 1 1

, ,

&c, each equal to 15 degrees. From A draw straight

lines through a, b, a1 ,
b1 ,
&c. These will be the hour
lines : A 9 c for 10 A. m., A a for
A. M., Aa. m., b for n
and so on. The proof of the correctness of this construc-
tion is easily seen by imagining the triangle A B C to be
turned round A B till it is perpendicular to the plane of
the paper or dial plate, and the triangle c D c 1 to be
turned up on c c till it abuts on B C, when D will coin-
1

cide with C, and A C will be parallel to the polar axis and

perpendicular to the plane of Dec
1
.

When the divisions on the line E B F run off the plate

we continue them thus : In Ac 1
(the 3 m. line) take
p.

any point 0, and through it draw a line parallel to A c, (the

9 a. m. line) meeting A b
1
A a 1 &c, ,
in
p p
1
&c, and
t ,

make q equal to op, q to op 1 1

, &c, and through #, q 1 ,

&c, draw straight lines A q, A q &c, which will be the

1

evening hour lines.

A similar construction serves for the morning hour lines
on the other side.

VERTICAL DIALS.
These have the advantage that they may be made of a
very large size and placed in conspicuous positions.
There are various ways of constructing them. As simple
a plan as any is to fix a flat disk, having a round hole in
it, in front of a wall with a southerly aspect. (Fig. 15.)
52 Vertical Sun Dials.

The disk should be roughly per-

pendicular to the sun's rays at
noon about the equinoxes. Tke^
The bright spot in the middle of
the shadow of the disk on the
wall indicates the hour. The
hour lines are found thus At :

the time the sun is on the meri-

dian mark the position of the
bright spot on the wall. Let A
be the hole in the disk and B
Fig- 15.
the spot. Measure A B. Through
B draw B C vertical, and draw a line B D so that B D is

equal to A B, and the angle C B D to the sun's polar

distance minus the co-latitude. Make the angle B D C
equal to the supplement of the sun's polar distance. It fol-

lows from this construction that if the triangle BCD were

turned round B C till it touched A the points D and A
would coincide, and C D (and therefore the imaginary
line C A) would be parallel to the polar axis. Now take
a watch, set to noon at the time of the sun's transit, and
mark the positions of the spot on the wall at the success-
ive hours. Straight lines joining these points with C will be
the hour lines.

Of course a large triangular stile CAB

might be
substituted for the disk or we might use a rod C A fixed
;

in the plane of the meridian, and having the angle A C B

(which it makes with the vertical)equal| to the co-latitude.
 CHAPTER VI.

REMARKS ON PORTABLE ASTRONOMICAL INSTRUMENTS.

THE REFRACTING TELESCOPE.
The refracting- telescope in its simplest form is a tube
with an object glass at one end and an eye piece at the
other. The former produces an inverted image of the
object at its focus in the same way as the lens of a
pho-
tographic camera, and the eye piece simply serves to
magnify this image. The eye piece usually has two lenses;
but if the telescope is wanted to show the object in its
natural position a combination of four lenses is employed,
by means of which the inverted image is again inverted.
This has, however, the disadvantage of cutting off too
much light, and is only used in small telescopes and for
land objects.

The ordinary eye piece is of two kinds first, the

;

"negative," in which the image is formed between the

two lenses of the eye piece. This is the kind generally
used in telescopes designed for the mere examination of
objects without making measurements. Secondly, the
"positive," in which the common focus of the two lenses
is outside the eye piece. This is the kind used in all
telescopes intended for measuring angles by means of
spider lines,&c. When the instrument is properly
focused the image and system of wires are in the same
 54 The Telescope.

plane at the common focus of the object glass and eye

piece. The
position of the focus of the former depends
on the distance of the object that of the latter on the
eye of the observer. The one is the same for every indi-
vidual. The other has to be adjusted to suit the observer
short-sighted people having to push the eye piece in,
while those who have
long sight require a longer focus.
The larger the object glass is the more rays from the
object are collected on the image, and the brighter it is.
The greater the magnifying power of the eye piece the
more apparent are any defects of definition in the image.

The magnifying power of the telescope is measured by

. focal length of object glass. _. ..
.

the cfraction . =-= ^. ? - r

2 r
Thus, if this
.

focal length of eye piece.

fraction were 4, the linear dimensions of the object seen

through the telescope would be four times what they

would be when viewed with the naked eye. Therefore,
for a given eye piece, the longer the telescope is the/^9^
smaller will be the field of view, or portion of the earth^ J "***
or sky visible. The angular diameter of the field is, in^y^
the angle subtended by the diameter of the eye -co^t,
fact,

piece at the centre of the object glass. **^y^ /,/^

^^ ^
In large telescopes the field of view is so small that it

necessary to use a 'finder," which

*
is is simply a small
telescope attached to it so that the axes of the two shall
be parallel.
A diagonal eye piece is one in which there is a mirror or
prism between its two lenses by which the rays of light
are turned at right angles and emerge from the side instead
of the end of the eye piece. It is used for observing

objects when the altitude is so great that it would be

uncomfortable or impossible to look up through the tele-
scope tube.
Lenses have to be corrected for chromatic aberration and
spherical aberration. Take the case of an object glass con-
 The Telescope. 55

sisting of a simple glass lens. Rays of light of different

colour having different degrees of refrangibility would
have different foci. The result would be indistinctness of
image, and, if the object were white, a sort of rainbow

appearance of coloured light. This chromatic aberration

is got rid of by using a compound lens consisting of two

lenses of glass of different refrangibility, one behind the

other. Such a lens is called "achromatic."

By "spherical aberration" is meant the dispersion of

rays caused by the central portion of a lens with spherical
surfaces having a different focus from its outer or marginal
part. This is corrected by the form given to the aber-
ration combination.

The way to ascertain the object glass has been

if

properly corrected for colour is to turn it on some bright

object, such as the moon or Jupiter, and put the eye
piece a little out of focus. This ought to produce either
a purple or pale green ring round the image, according as
the lens is pushed in or drawn out.

Spherical aberration is detected by covering the central

portion of the lens with a circular disk of paper and
focusing it on an object, afterwards removing the disk

and covering the outer part with a ring-shaped piece, when

the focus ought to remain the same.

If one part of the object glass has a different refractive

power from another part a bright star will appear with an
irradiation, or wing, at one side.

THE MICROMETER.
The micrometer is a contrivance for measuring small
angular distances. It is, like the cross wires of an ordi-

nary theodolite, placed in the common focus of the object

glass and eye piece of a telescope, or of the "reading
microscope," which will be described presently. The
principle of the micrometer is simply this :
Suppose that
56 The Micrometer.

a point such as the intersection of the cross wires can

be moved across the field of view by means of a screw.
Let a be the angular diameter of the field, and let n be the
number of complete turns of the screw required to move
the wires through this space ;
it is evident that one turn

of the screw will move them through an angle

The head of the screw should of course have an index to

mark the commencement and end of each turn. If the
head is made large enough to enable its rim to be divided
into m equal parts we shall have the means of measuring

an angle by turning the head through one division.

Thus, by making the thread of the screw sufficiently fine,

and its head large enough, we have the means of measur-
ing small angles to an Extreme degree of accuracy, pro-
vided we know the angular value of one turn of the screw.
This may be ascertained by finding how many turns it
takes to move the wire across the image of an object of
known dimensions at a known distance. A levelling rod
will answer the purpose. The length of rod moved over,
divided by the distance, gives, of course, the chord of the
angle subtended. There is usually a scale in the field of
view, the divisions of which correspond to the turns of
the screw.

There are several different forms of micrometer. The

accompanying figure (16) represents the one known as the
filar (or thread) micrometer. Two parallel wires m w,
 The Micrometer. 57

n n, are fixed to two frames, a, 6, sliding one within the

other, which are moved by screws as shown. The frame
b slides within an outer fixed frame c. The frame a carries
the wire m m, and is moved by the screw A, the head of
which has a graduated rim. b carries the wire nn and is
moved by the screw B. In the oval opening of a is a hori-
zontal wire at right angles to the others. Suppose we
have to measure the angular distance between two stars.
The instrument must be turned till both are on the hori-
zontal wire. If now the screw A is turned till mm cuts
one star, and B till n n cuts the other, the distance be-
tween the two is measured by the number of turns of A
and fractional divisions of a turn it takes to bring m m up
to n n. This is not the exact method of procedure followed,
but it serves to illustrate the principle.

THE READING MICROSCOPE.

The reading microscope is really a small telescope with
a micrometer in its focus, and is used instead of a vernier
for reading the fractional parts of the divisions of the

graduated circles of large instruments. The microscope

is fixed, the circle being attached to the telescope of the

instrument and moving with it. The micrometer has only

one screw and moveable frame, the latter carrying a pair
of cross-wires in the common focus of the object glass
and eye piece of the microscope. These cross-wires are
used in exactly the same way as those of a theodolite
telescope, only that the object viewed is the graduated
arc, on which the microscope must, of course, be focused.
To make the matter clear we will take the case of the
measurement of a horizontal angle by a large theodolite
the arc of which is graduated to 10 minutes. Suppose
that one turn of the micrometer screw is equivalent to one
minute, and that its head is divided into 60 parts. We
have thus the means of measuring the angle to single
seconds. The circle with its attached telescope is re-
volved, and the cross-wires of the latter made to coincide
58 The Reading Microscope.

with one of the objects. On viewing the arc through the

microscope (which it must be remembered is a fixture)
the wire intersection of the latter must be made to coin-
cide with one of the divisions of the arc by means of the
micrometer screw, and the reading of the index of the
latter noted. Suppose the arc reading to be io 20', and
that of the screw head 15". Now move the circle and
bring the telescope to bear on the other object. The
cross-wires of the microscope will probably fall some-
where between two divisions of the arc, say between 50
30' and 50* 40'. Turn the screw till the cross-wire is on
the 50" 30' division, and suppose that it takes between
three and four turns, and that the index marks 25". The
micrometer wire will have been moved 3' 10", and the true
reading of the second object will be 50* 33 10". The
angle measured is therefore 40 13' 10".
Two or more reading microscopes are placed at equal
distances round the circle, and the readings of all taken.
Errors due to eccentricity are thus got rid of, and those
due to faulty graduation and observation much diminished.

THE SPIRIT LEVEL.

The spirit level is used, not only to bring certain lines

of an instrument as nearly as possible into a horizontal
position, but also to measure the deviation of these lines
from the horizontal. For thispurpose the glass tube is
graduated, usually from its middle towards both ends,
and the reading of the ends of the air bubble noted. The
length of the bubble depends upon the temperature, and
the latter should therefore be also noted.

To obtain the value of one division of the level that is,

of the vertical angle through which the level must be
moved in order that the ends of the bubble may be dis-
placed one division a simple plan is to rest the level
upon some support (such as the horizontal plate of a
theodolite) that can be moved vertically and which is con-
 The Spirit Level. 59

nected with a telescope. The plate is levelled, the tele-

scope directed on a vertical measuring rod set up at a
known distance, and readings taken of the ends of the
bubble and of the intersection of the horizontal wire with
the rod. The whole arrangement is then moved vertically
by means of the foot screws till the ends of the bubble have
moved a certain number of divisions say 10. The read-
ing of the telescope wire on the rod is now noted. The
difference of the rod readings, divided by its distance,

gives the chord of the vertical angle moved by the level,

and this divided by 10 gives the value of one division.
In the case of the level of the transit telescope at the
Royal Military College it was found that at a distance of

383 feet a vertical movement that displaced the bubble

20 divisions altered the reading on a levelling staff 0.24
feet. The resulting decimal gave 0.0000313 as the
tangent of the subtended angle for one division, which
made the value of the latter 6". 45.
In the case of a striding level, if its feet, or the surface
on which they rest, are truly horizontal, while one leg is

longer than the other, the bubble reading of the end which
ishighest will be greatest, and if the level is turned end
for end the bubble readingswill change places. On the
other hand, the legs are of equal length, but the surface
if

on which they rest is not level, the bubble readings will

remain the same on reversal. That is, if we call one end
of the level A and the other B, the level error will make
the reading of A the largest in both positions. If a truly

adjusted level rests on a slope, and A reading is the

greatest in one position, B will be greatest by the same
amount on reversal.

In practice we generally find both errors combined,

especially in the case of the more delicate levels which
easily get out of adjustment. The amount of slope of the
surface tested by the level is obtained thus: Take the
6o The Spirit Level.

case of the pivots of a transit telescope. Placing the level

upon them take the readings of the bubble ends, and call
the reading next the west pivot and the other E. W
Then reversing the level take the readings over again and
call them 1
W
and E 1 The number of divisions by which
.

the bubble isdisplaced by the difference of level of the

pivots is given by the formula :

W+W 1
(E+E*)
4
find the actual slope of the pivots we must multiply
To
this quantity by the value of a division of the level.

The level error is obtained separately by simply chang-

ing the signs of W and E the above formula, when
1 1
in

we have:
Level error =
W W E + E = W-E (W Ei) 1 1

J
1

4 4
Of course if W+W 1 = E -f E
slope, and, in
1
there is no
practice, when out of adjustment, we may get
the level is

the points of support horizontal by raising one of them

till this is the case: For instance; if were 20 and E W
10, W 1
would have to be 10 and E 1
20.

If the level is in adjustment we must have

W E^W 1
E 1

In this case we have only to take W and E and the

slope is obtained from the formula

X value of one division

2

Example Take W=25, E = io, W J = i5, E 1

=20.
Value of one division =6".
25
Here we have--
, + 15 10 20
= 10
=2^2
4 4
Multiplying this by 6" we have 15" as the slope, the west
pivot being the highest.

The 25
~ I5 ~ IQ+2 o
level error is x6"= 3 o //

4
 The Chronometer. 61

If great accuracy is required the level should be read a

number of times in each position, lifting it up after each
reading and taking the same number of readings in the
two positions. The difference of the sums of all the read-
ings at the two ends divided by the total number of the
readings will give the slope.
The value of a division of the level should be ascer-
tained for different temperatures.

To correct the error of the level get the surface on

which it rests truly horizontal as explained above.
Then,
by means of the adjusting screw, move the bubble till
both its ends read the same.
THE CHRONOMETER.
The chronometer is simply a very perfect watch with-
out a regulator, and with the balance so constructed that
changes of temperature have the least possible effect upon
the time of its oscillation. Chronometers may be con-
structed to keep either mean or sidereal time. Those
used on board British ships are intended to show Green-
wich mean time. The great point in a chronometer is
that it should keep a regular rate ;
that is, that it should
only gain or lose a certain amount in a given time. If
this can be depended on we can always ascertain
the true time at any instant by applying the rate for the
number of days and hours that have elapsed since the actual
error of the chronometer was last determined, whether

by comparison with other chronometers or by an astrono-

mical observation. The more regular the rate kept the
more perfect is the chronometer. It is also more con-

venient to have a small than a large rate to allow for.

Chronometers are generally made to run either two or

eight days. The former are wound
daily, the latter every
seventh day. It is important that they should be wound
up at the regular intervals, as, if let to go too long, an
unused part of the spring comes into play, and irregu-
62 The Electro Chronograph.

larity of rate may result. If a chronometer has run down

it requires a quick rotatory movement to start it after it

has been wound.

Transporting On
board ship chronometers are allowed
to swing freely in their gimbals so that they may keep a
horizontal position but on land they should be fastened
;

with a clamp. Pocket chronometers should always be

kept in the same position, and if carried in the pocket in
the day should be hung up at night.
Chronometers have usually a different rate when
travelling from what they keep when stationary. The
travelling rate may be found by comparing observations
for time taken at the same place before and after a
journey, or from observations at two places of which the
difference of longitude is known.
For mean time observations an ordinary watch may be
used by comparing it with the chronometer, provided the
rate of the watch is known.
Chronometers are generally made to beat half seconds.

THE ELECTRO CHRONOGRAPH.

Under head may be included all contrivances for
this

registering small intervals of time by visible marks pro-

duced by an electro magnet, and thus recording to a
precise fraction of a second the actual instant of an
occurrance. By this means an observer at a station A
can record at a distant station B the exact instant at
which a given star passes his meridian, and thus the
difference of longitude of the two stations may be ascer-
tained.

REFLECTING INSTRUMENTS.
THE SEXTANT.
Aperson accustomed to work with the pocket sextant
will have little difficulty in using the larger kind and the ;

latter, with its adjustments, is so fully described in most

 The Sextant. 63

works on surveying that little need be said about it here.

On land it is chiefly used for taking altitudes with an
artificial horizon. The latter usually consists of a trough
of mercury with a glass roof to protect it from the wind,
or with a plate of glass floating on the mercury. The
trough should be a sufficiently large one, otherwise the
observer will be continually losing time and patience by
failing to catch both images together. To eliminate
errors caused by want of parallelism in the glass of the
roof, when one half of a set of observations has been taken
the roof should be reversed end for end. For taking the
sun's double altitude the dark glass of the eye piece may
be used. The two images should be made to overlap
each other a little, the vernier clamped, and the instant
noted when the circles just touch. As this requires that
the images should be receding from each other, the alti-
tude of the lower limb must be taken in the forenoon and
of the upper limb in the afternoon. For a lunar distance
of the sun direct the telescope on the moon and use one
or more of the hinged dark glasses for the sun. The
index error should be obtained, and applied as a constant
correction.
A common fault of the sextant is that the optical power
of the telescope is too small. There is little use in being
able to read the graduation to jo seconds if the eye can-
not be sure of the contact of the images within 30".
THE SIMPLE REFLECTING CIRCLE.
This is simply a sextant with its arc graduated for the
whole circumference of a circle, and with the index arm pro-
duced to meet it at opposite points and carrying a vernier
at each end. The mean of the two verniers can be taken
at each observation and any error due to eccentricity thus

got rid of. This arrangement also tends to diminish

the
errors of graduation and observation.
Some reflecting circles have three verniers at intervals
of 120.
64 The Repeating Reflecting Circle.

THE REPEATING REFLECTING CIRCLE.

In the repeating reflecting circle the horizon glass
(m Fig. 17), instead
of being immovable,
is attached to an
arm which revolves
about the centre of
the instrument and
which also carries
the telescope (/) and
a vernier (v). Jhe
7V1
index glass (m) is

carried on another
revolving arm, which also has a vernier The arc is
graduated from o to 720 in the direction of the hands of
a clock. To use the instrument the index arm is clamped
and its reading taken. The telescope is then directed on
the right hand object (6), the circle revolved till the
images coincide, and the telescope arm clamped. The
index arm then undamped, the telescope directed on
is

the left hand object (0), and the index moved forward till
the images again coincide, when its vernier is read. The
difference between the two readings of the index vernier
is twice the angle between the objects. This repeating
process may be carried on for any even number of times.
The first and lastreadings only are taken and their
difference, divided by the number of repetitions, gives the
angle. If the angle is changing, as in the case of an alti-
tude, the result will be the mean of the angles observed,
and the time of each observation having been noted the
mean of the times is taken.

This instrument will not measure a greater angle than

the sextant. advantages over the latter are that there
Its
is no index error, and errors of reading, graduation, and
eccentricity are all nearly eliminated by taking a sufficient
number of cross-observations.

#^ / 7 y^/r<^5 4o rt*. oUA xVt/M- c*-0^-'UL/l

 The Prismatic Reflecting Circle. 65

The dip-measurer a repeating circle which has a third

is

mirror (n) mounted on the telescope arm at an angle of

about 45 to the horizon glass, but only one-half the
height of its silvered portion. By means of this modifi-
cation angles of even more than 180 can be measured,
and the instrument can therefore be used for taking double
altitudes of objects near the zenith.

THE PRISMATIC REFLECTING CIRCLE.

In the prismatic reflecting circle the telescope (t) is
fixed, and instead of the horizon mirror there is a small
fixed prismatic reflector (p)
which half covers the object
glass. The index mirror
(m) is carriedon an arm
which revolves round the
centre of the circle and has
a vernier (v v1) at both
ends. This instrument will
measure angles of any di-
mension, and has also the
following advantages: (1)

Eccentricity is completely
eliminated by using both
Fig. 18.
verniers. (2) The reflected

images are brighter than in the case of other reflect-

(3) The prismatic reflector
ing instruments. causes less
error than the ordinary one.
At angles about 180 the observer's head gets in the
way, when a prismatic eye piece is used.
The prismatic sextant differs from the circle in having
the whole
only one vernier, and in the arc not extending
circumference. It measures the same angles as the circle,
but does not eliminate the eccentricity.
(For a full description of the above reflecting instru-
ments, with their adjustments and the method of using
them, vide Chauvenet's Astronomy.)
 CHAPTER VII.

THE PORTABLE TRANSIT INSTRUMENT.

The transit instrument is a telescope with two trunnions

resting on Y-shaped supports so that its line of collima-

tion may move in a vertical plane, and is used for the

purpose of taking the times of transit of heavenly bodies

across that plane generally either the meridian or the
prime vertical. In the former case it enables us to find
the true local time, either mean or sidereal, and also
serves to determine the longitude by means of transits of
moon-culminating stars. In the latter case it gives us a
very accurate method of ascertaining the latitude by
transits of stars across the prime vertical.

In the focus of the telescope are one or two horizontal

wires, and an odd number of equi-distant vertical ones
generally five of which the central should be in the
opti-
cal axis of the instrument, and at right angles to the axis
of the trunnions or pivots; and if, in addition, this axis is
truly horizontal, the line of collimation will move in a ver-
ticalplane. Thetelescope provided with a vertical
is

graduated circle, with a level attached, which serves as a

finder to set it at any required angle of elevation. It has
also a diagonal eye piece for transits of objects of consid-
erable elevation, and a very delicate striding level for
getting the pivots perfectly horizontal. At night the light
 The Transit Telescope. 67

of a lantern is thrown into the interior to illuminate the

wires by means of an opening with a lens in one of the
pivots and a small internal reflector. In taking transits it
is usual to note the time of the object's passing each wire,

and to use the mean of all the wires instead of the transit
across the central one. In the field the best plan is for

an assistant to hold the watch or chronometer, the ob-

server calling out "stop" as the object passes each wire.
In the case of the sun, moon, and planets the instant
noted is when the edgeof the object touches the wire,
either in coming up or leaving it, the time required
to it

for its semi-diameter to pass the meridian being after-

wards added or subtracted, z/^/^w bof^ lim&s <-&-&
o6setriy&
7,71, Tvfcrj&Ti. rrrsc fJvt*. /->f-e */// / <s fjcejfa&n

The first adjustment to be attended to is that of colli-

mation. This may be effected by getting the central wire
on some well-defined distant object, or on a circumpolar
star at its greatest elongation. The telescope is then re-
versed in its supports, end for end, when, if the wire still
bisects the object, the collimation is all right. If it falls to
one side of it it must be moved towards it half the inter-
val by the collimation screws. The instrument is then
moved laterally by means of small screws connected with
one of the Y supports till the wire bisects the object,
when the telescope again reversed and the process re-
is

peated till the collimation is perfect.

The horizontality of the axis of the pivots is obtained

by the striding level and foot screws. As the level has
generally an error of its own which is itself liable to
change (owing to alterations in temperature, accidental

flexure, &c.) be found convenient to level the

it will

pivots by getting them in such a position that the level

will have equal but opposite readings in reversed posi-
tions. Thus, if in one position the east end of the bubble
reads 10, while the west end reads 12, then, on reversal, the
east end should read 12 and the west 10.
68 The Transit Telescope.

one of the pivots has a larger diameter than the

If
other it is evident that when their
upper surface is level
their axis will not be so. This will entail a constant
error which will be investigated presently.

The verticality of the central wire must be tested by

levelling the pivots and noticing whether the wire re-
mains upon the same point throughout its whole length
when the telescope is slowly moved in altitude.
If the collimation is out of adjustment, but the level-
ling correct, the of collimation will sweep out a
line
cone. If the collimation is correct but the levelling in-

accurate, it will describe a great circle, but not a ver-

tical one. If both are right it will move in a vertical

plane. We
have now to make this plane coincide with
some given one say that of the meridian. The north
and south line may have been already approximately ob-
tained by means of a theodolite, and we can now find it
exactly by one of the following methods.

(i) By transits of two stars differing little in right as-

cension, one as near the pole, the other as far from it as

possible. Let a be the right ascension, ft the declination
and t the observed clock time of transit of the star near
the pole; a 1 ft 1 and t the same quantities for the other
1
, ,

star, d the azimuth of the instrument in other words,

the error or deviation to be determined and f the lati-

tude. Then d is found from the formula,

*; ; -(*;-h cos a cos*

-*) <** $1 cos 1
'={^L <p
sin (ft ft )

The must be known, but not its error;

rate of the clock
the interval must be corrected fcr error of rate; and, if
t
1
t

a mean time watch is used, converted into sidereal time.

d being in horary units must be multiplied by 15 to ob-
tain the error in arc.

If the declination of the southern star is south it will.

 The Transit Telescope. 69

of course, be negative in the northern hemisphere, and

vice versa. ,
,
.

If the i nt e rval between-4he -tra n s i ts is-feeo-smail the^^f^ ^^.

deviation will be west of north, and vice versa. In the
southern hemisphere the contrary would be the case.

(2) By observing the transits of two circumpolar

stars which culminate within a short time of each other
one above, the other below the pole, d Ursce Minoris and
51 Cephei are a good pair for this. These stars being
very slow moving it is evident that a slight deviation of
the instrument will cause one to come to the central wire
too soon, and the other too late. The formula is

^ _A^ ^
2
A ^:- t
l-^ ^! H
Ci
)
COS d COS fil ffr
{ J
cos <p
sin (O +O 1
)
<0>^^ s*Jo*j.

The value of the fractional part of this formula at Kings-

ton (Lat. 44 14') is 0.0372 for the above pair of stars.

If the interval between th / two transits is too great,

<

the deviation is west of north and vice versa. It is always

safest to draw a figure in these cases.
The deviation having been calculated the position of
the meridian mark can be corrected if its distance from
the instrument is known.
To prove the formula:
cos d cos d 1
j
d=
f

J
(
,
a1 a ,..
(^ m
*.|

cos
;

sin (a
^ ir-r-
]
)
J (p

Let P be the pole, Z the zenith,

A 1 Z B 1 the plane in which the tele-
scope moves, and A Z JSthe true meri-
dian. We have to find the angle
AZ A 1
.

Let S and S 1 be the two stars at

transit, e the unknown clock error, /

the clock time when the star S was on Fig. 19.

70 The Transit Telescope.

the meridian. The true time of the star being at S will

be t+e.
Let a be the R. A. of .the star. Then a was the time
when the star was on PB; Z P S t + e a. .*. =
Let <p be the observer's latitude, d the star's declina-
tion, d the deviation of the instrument.
From the triangle P Z S we have :

Sin S Z sin S Z B = Sin P S sin S P Z (i)

And Z S = V S P Z,very nearly=y>-
/. sin (<p d) sin d sin {t-\-e a) cos d
or (sin <p
cos d cos <p sin d) d=(t-\-e a) cos d
or (sin <p
cos <p
tan 3) d=t + e a (2)
If a 3 l be corresponding quantities for star S 1 we have
1
,

(sin <p cos <p tan f. 1 ) d t x +e a 1 (3)

In equation (3) t 1 includes the correction for the clock
rate between the observations.

Subtracting (2) from (3) we have

d cos <p (tan tan 8) = t*t{a f
a) (4)
sin
tan oy =
(<5
Now
XT
tan 5
<?')
r>
cos cos
.J j- Li ^ / '
\1 cos <* cos $'
;. d.=-#r t * (a a) .

I J
cos p. sin {6 6'
It is evident from equation (4) that for a given value of
1
d the quantity t t'ia
1
a) is larger as tan d tan tf'is
larger. In other words, one of the stars should be as near
the pole and the other as far south as possible.
Equation (1) may be put in the form
c Z7r,c d sin Z S
Sin P S = ,
cos
As the angle Z P S is the error, in time, of transit,
caused by the azimuthal deviation d, this equation gives us
the means of correcting a transit where it has not been
convenient to correct the meridian mark.
TO FIND THE ERROR DUE TO INEQUALITY OF PIVOTS IN
THE TRANSIT TELESCOPE.
In Fig. 20 let A C, B D, be the diameters of two un-

equal pivots, E F their axis. If the side A B on which

the feet of the level rest is horizontal, the lower side
 Inequality of Pivots. 7*

C D be inclined at a certain angle, say

will a, and EF
a
will be inclined at an angle
<4i$. ~2-
If, now, the instrument be
reversed in the Y's, A B will

evidently be inclined at an
angle 2 a, which
be given
will

by the level readings in the usual way.

Therefore, the inclination of the axis in the first position
will be one quarter that given by the difference of the
inclinations given by the level readings in the two positions.
This quantity will be a constant correction to be applied
to the inclination of the pivots as given by the level.
For instance, in the case of the instrument at the Royal
Military College the finder pivot was fonnd to be the
thickest,moving the bubble 4 divisions on reversal.
Therefore, when the upper surface of the pivots was level
their lower surface was inclined for two divisions, and
their axis for one. The level error has, therefore, to be
corrected for the value of one division; e. g. with west t

pivot highest by three divisions and finder east, the total

error will be four. It reduces the error when the finder
pivot is highest, and vice versa.

TO APPLY THE LEVEL CORRECTION TO AN OBSERVATION

WITH THE TRANSIT TELESCOPE.
The correction having been found in the usual
level

way, be the pole, Z the zenith, S the star when on

let P
the central wire, R the south point of the horizon.
P R S is a spherical triangle in which P and R
are very small angles. R is the level correction.
M
OS
XT
Now -=r.
Sin PS = Sin
-^
S R
^
jM Sin R
_5
bin P
Iv^l
or cos. declination sin. altitude
R
R sin. altitude
and P ==
cos. declination
72 The Level Correction.

p
The correction for the transit in time will be
15

Example At Kingston, Canada, the transit of Arcturus

was observed, the level readings being :

1st position East 45 West 35

"
2nd East 25 West 55
To find the correction in time.

Here we have
Latitude 44* 14' N
Star's declination 19 47 N
Star's altitude at transit 65 33
S5~f J55 45 2S 20
Level correction s=
OJ ^ -^ J = 5 divisions,
,. . .

4 4
west end being highest, and the pivot correction altered
this to 6 divisions. The value of one division of the level
was 6". 45, therefore the angle R was 38". 7 east of the

meridian, and the transit took place too soon.

P =
\\ x 38". 7 = 37". 2, and the correction was
2.48s. to be added to the observed time of transit.

When the instrument is in perfect adjustment the error

of the watch or chronometer can be at once obtained by
means of meridian transits, as described at page 34.

FINDING THE LATITUDE BY TRANSITS OF STARS ACROSS

THE PRIME VERTICAL.
If S is a star on the prime vertical, P the pole, Z
the zenith, and W Z E the prime
vertical,S P Z is a right-angled
triangle and if we know the
;

angle S P Z and the side P S we

can find the side P Z by the
Fis 22 -

equation
Cos S P Z=tan P Z cot P S
 Prime Vertical Transits. 73

or, if a is the star's right ascension, ^ the sidereal time of

its crossing the prime vertical, d its declination, and
<p
the latitude of the place.
Cds ( t)=cot <p
tan d
If the star is in the position S' (or west of the meridian)
the equation becomes

Cos {t a) = cot <p

tan d
This method gives the latitude very accurately, pro-
vided the true sidereal time of the transit is known. The
observation may be made either with a portable transit
instrument or with a transit theodolite. In this, as in
other delicate methods, the latitude is supposed to be
approximately known beforehand. We can then from
the above equation calculate the approximate time of the
transit. The star's altitude is found from the equation

Sin. altitude =
sin <p

(Since cos P S = cos PZ cos Z S.)

It is evident that no stars cross the prime vertical

except those whose declinations lie between zero and the
latitude of the place. Those which culminate near the
zenith are to be preferred, because a small error in the
observed time of transitwill affect the result less.

is used the true meridian line

If the transit theodolite

may be obtained by any of the ordinary methods, and the

telescope clamped at right angles to it by means of the
horizontal arc. If we are working with the portable
transit telescope the method adopted is to calculate

(from the approximate latitude) the time at which a star

which culminates several degrees south of the zenith will
cross the prime vertical, and direct the telescope on it at
that instant. It will now be nearly in the required

position. The error may be determined as follows:

Take the sidereal time of transit of a star over both the
74 Prime Vertical Transits.

east and the west verticals. The mean of the two will
be the time of the transit over the meridian of the instru-
ment, and should be equal to the right ascension of the
star. If the two results are not equal their difference
shows the angle which the plane of the instrument makes
with the true prime vertical.

In working these observations we may use either a

sidereal or amean time chronometer, in the latter case
making the usual reductions, and always allowing for the
rate. If two transit telescopes are available, one of
them may be upset plane of the meridian for the
in the

purpose of ascertaining the exact chronometer or watch

error by star transits. A large transit theodolite serves
instead of two transit instruments, and in this case an
ordinary good mean time watch will suffice, the mean time
of the observations being reduced to sidereal time. If
both the east and the west transits are observed the dif-
ference of time in sidereal units is double the hour angle
P, and the latter may therefore be obtained without any
reference to the actual watch error, provided the rate is
known. It should also be noted that if we reverse the
telescope on its supports any error of collimation or ine-
quality of pivots will produce exactly contrary effects on
the determination of the latitude. Two stars may be
observed with the telescope in reversed positions on the
same day, or the same star on two successive days, and
the mean of the two resulting latitudes taken.
It will be found advisable to calculate beforehand the
altitudes and times of transit (either mean or sidereal, as
the case may be) of a number of suitable stars.

If the plane of the telescope is not in the prime vertical

the calculated latitude will be too great. Suppose the
deviation to be to the east of north and that the tele-
 Prime Vertical Transits.
75

cope describes a vertical circle

passing through th^E
1
Z W 1
.

Then P Z 1
,
which bisects
S S1 ,
will be the calculated
co-latitude. The correction
for the may be
deviation
computed The thus.
star's
R. A., minus the mean of the Fig. 23.
times of transit corrected for clock error, will be the angle
ZPZ 1
Now, from the right-angled triangle Z P Z 1
.
,

we have :

tan PZ cos ZPZ' = tan PZ' = tan PS cos SPZ'

tan d cos ZPZ'

or tan <p
cos SPZ 1
angle ZPZ is the same for all stars, and it is bet-
The 1

ter to obtain its value from a star which culminates

several degrees south of the zenith, since the same error
in the observations will have less effect upon the calcu-
lated azimuth.

If the intervals between the vertical wires are not all

exactly the same a correction has to be applied. For
details on this point vide Chauvenet.
In the field generally mounted on a
the instrument is

wooden portable stand with three legs. In order to obvi-

ate the effects of vibration produced by the observer's
movements the ends of the legs may be made to rest in
notches in flat blocks of wood placed at the bottom of

holes dug in the ground about eighteen inches deep and

two feet in diameter. This was the plan adopted on the
North American Boundary Survey between the Lake of
the Wocds and the Rocky Mountains.
The meridian mark should, if possible, be at least half
a mile distant. A black or white vertical stripe painted
on a stone serves for the day time. At night a bull's eye
j6 The Personal Equation.

lantern may be used, the glass being covered by a piece of

tin with a vertical slit cut in it. Or, as the lantern is
liable to be blown out by the wind, it may be enclosed in
a wooden box with a vertical slit.

The larger transit theodolites may be used as transit

instruments, and have the advantage over them that when
the meridian line has been ascertained the prime vertical
can be at once set off.

THE PERSONAL EQUATION.

It often happens that two persons, equally well trained
in taking observations, will differ by a considerable and
nearly constant quantity in estimating the precise instant
of an event, such as the transit of a star across a wire.
This difference is called their personal equation, and an
allowance should always be made for it when observations
made by two individuals have to be combined. In the
case of the transit instrument this equation may be de-
termined as follows: Let one observer note the passage
of a star over the first three wires and the other observer
note the transits over the remaining wires. If the two
observers' estimation of the instant of transit differ, it is

evident that (provided the wires are equidistant) the

difference will appear on comparing the intervals of time.
For instance, if A notes the transits across the first three
wires at ios., 20s., and 30s., and B
notes the remaining
two at 39S.5 and 49s. 5, it is plain that A would consider
the star to be on any wire half a second later than B
would, and their personal equation is therefore os.5. By
repeating the same process on other stars, and taking the
mean of the result, a more accurate estimate is obtained.
The personal equation has been found liable to vary with
the state of health of the individual.

The difference in the estimated instant of a transit is

only a particular case of the personal equation.

 CHAPTER VIII

THE ZENITH TELESCOPE.

The zenith telescope is a contrivance for the exact de-
termination of the latitude by measuring with the greatest
minuteness the differences of the meridian zenith dis-
tances of two stars, one of which culminates north and
the other south of the zenith, within a short interval of
time of each other.

Let X be the latitude, d and z the declination and

zenith distance of the southern star, d' and /those of the
northern star. Then, since the latitude is the same as
the zenith distance of the equator, we have :

l=d + z
X=$z
and adding, 2A=d+&-\-zz'
Therefore, if d and d> are known exactly we can find
the latitude from the difference of z and z without know-
ing their actual values. Moreover, if z and z are nearly
the sams the refractions will nearly neutralize each other,
and we shall only have to take into account the difference
of the refractions at the two altitudes.

The instrument is practically a telescope about 45 inches

focal length, attached to a vertical axis round which it

revolves, having been first clamped at a certain angle of

78 The Zenith Telescope.

elevation. The latitude must be known approximately,

and a pair of stars selected which are of so nearly the
same meridian zenith distance at that latitude that they
will both pass within the field of view of the telescope
without our having to alter its angle of elevation. As a
rule, x and z must not differ by more than 50' at the most.
If the axis is truly vertical and the telescope remains at
the same vertical angle at the observation of both stars,
then it is plain that the difference of z and z may be read
by a micrometer in the eye piece.
It is usual to observe only stars which pass within 25

degrees of the zenith. The telescope has a long diagonal

eye piece with a micrometer in its focus, and the micro-
meter wire is at right angles to the meridian. There is a
very delicate level attached to the telescope, and a vertical
arc which serves as a finder. By reading this level at
each observation we can detect and allow for any change
in the angle of elevation of the telescope.
The above is the merest outline of the principle
of the instrument, and reference must be made to
other works for the details of its construction. The
method of using it is this The latitude being already
:

approximately known, a pair of stars is found from a star

catalogue, both of which will pass within the field of view
without altering the elevation, and which have nearly the
same right ascension. The reason for this is that their
transit may take place within so short an interval of time
that the state of the instrument may remain unchanged ;

but a sufficient interval must be allowed for reading the

micrometer and level and reversing in azimuth say, not ;

less than one minute or more than twenty. The meridian

line must have been previously ascertained by transits of
known stars, or otherwise, and the chronometer time
calculated at which each of the stars will culminate. The
telescope having been broughl into the meridian, ready
for the star which culminates first, and set for the mean
 The Zenith Telescope. 79

altitude of the two stars, the observer watches the first star
enter the field of view, and bisects it with the micrometer
wire at the calculated instant of its transit. He then reads
the level and micrometer, revolves the telescope in azimuth
through 180 and observes the second star in the same
,

manner. If, after the revolution, the level is much out, it

must be relevelled by a tangent screw provided for the
purpose, which does not alter the connection between the
telescope and the level, but moves them together.
The micrometer has a single screw and one or more
moveable wires, and reads to as little as o".05. It usually
has several fixed vertical wires, so that the instrument
may be used for transits.
Besides the vertical arc for setting the telescope at the
required elevation, there is a horizontal circle with two
stops for getting it in the plane of the meridian, and also
a striding level for setting the axis vertical.
In the field the instrument is mounted like the transit

telescope.
This method of finding the latitude is known as Tal-
cott's,having been invented by Captain Talcott of the
U. S. Engineers. Its defects are that it is often difficult
to obtain a sufficient number of suitable pairs of stars, of
which the declinations are accurately known. As a rule
we have to use the smaller stars, whose places are not
very well known, and must therefore observe a large num-
ber of pairs to eliminate errors.
TO FIND THE CORRECTED LATITUDE.
Let m
be the micrometer reading (in arc) for the south-
ern star, m the same for any point in the field assumed as
the micrometer zero, and z the apparent zenith distance
represented by m when the level reading is zero. Suppose,
also, that the micrometer readings increase as the zenith
distances decrease. Then, if the level reading were zero,
the star's apparent zenith distance would be
8o The Zenith Telescope.

z +m 7)i

Letbe the equivalent in arc of the level reading, posi-

/

tion when the reading of the north end of the level is the

greater. Let r be the refraction. Then the true zenith

distance of the southern star, or z, is:

~o+ w o m + l-\-r
The quantity z Q +m
is constant so long as the relation

of the level and telescope is not changed. have, there- We

fore, for the northern star,
z =z -\-m m' /' +r
Hence
z z'=m m + l'-\-l + r f
and the equation for the latitude previously given will
become
W W)
:

+b (m -m)+i (l'-tl) +i (r-'S)

TO FIND THE CORRECTION FOR LEVEL.
Calling the readings of the north and south ends of the
bubble n and s, and the inclinations at the observations of
the north and south stars, expressed in divisions of the
level, L' and L, we shall have

2 2
and if D is the value of a division of the level in seconds
of arc, we have
r=L' D /=L D
and the correction for the level will be

i
(/'+/)=! (L'+L) D= n'+ n (s '+ s) D
4
TO FIND THE VALUE OF A DIVISION OF THE LEVEL.
Turn the telescope on a well-defined distant mark.
Set the level to an extreme reading L, bisect it by the
micrometer wire, and let the micrometer reading be M.
Now move the telescope and level together by the tangent
 The Zenith Telescope. 81

screw till the bubble gives a reading L/ near the other ex-
treme, bisect the mark again by the wire, and let the
micrometer reading be M'. The value of a division of
the level in turns of the micrometer will be
.
imm
M-M'
U=t
and if R is the value in seconds of arc of a revolution
of the micrometer, the value D of the level in seconds of
arc will be
D=Rd
TO FIND THE VALUE OF A REVOLUTION OF THE MICRO-
METER SCREW.
This is best done by observations of a circumpolar star
near its
greatest elongation. find We its hour angle and
zenith distance by the formulae
cos t=cot d tan X
cos 2=cosec S sin X
Whence, knowing the star's R. A. and the chronometer
error, we find the chronometer time of the greatest elon-
gation. Set the telescope for the zenith distance z, direct
it upon the star 20 or 30 minutes before the time of great-

est elongation, and bisect it with the micrometer wire;

note the time of bisection, and micrometer and level read-
ings. As the star moves vertically repeat this process as
often as possible while it is in the field of view. Let
^i> ^2 t$ > &c-> be the noted chronometer times of bisec-
tion, Wj, w a m 5 &c,
, ,
the corresponding micrometer
readings, m the micrometer reading at the instant of
greatest elongation (/), and i lt i a , i z &c, the required ,

angular distances then the latter are found from the equa-
;

tions
sin *'
1 =sin (t t
x)
cos S
sin *'
3 =sin (t 1
9) cos d

If Rthe value, in seconds, of a revolution of the

is

micrometer, and if the level has remained constant, we

have, since (m m t)
is the number of turns given to the
82 The Zenith Telescope.

screw to move the thread through the angular distance ij

(m Wj) R=i t

Also (m w 2 R=iz )

Therefore, subtracting

m% m 1

Tocorrect for any change in the level reading, let l t

and / a be the level readings corresponding to x and ;;z 3 m ;

then (/ 3 /
t )
D is the change required. The angular value
of D is unknown ; but, since T)=dR, the correction to be

applied to (i t i2) is (l 2 h) dR; and

(w, m x)
R=i 1 i% (/ 3 /j) dR

(m a m t) =f (/ 3 l
t)
d
A value of R is thus obtained for each of the observa-
tions, and the mean of the results taken. This mean has
then to be corrected for refraction, thus : From the tables
find the change in refraction for i' at the zenith distance
z. Let this change be dr then ;
R dr will be the correc-
tion to be subtracted from R.

REDUCTION TO THE MERIDIAN.

has not been observed exactly on the meri-
If a star
dian may be taken when off it, and the observation re-
it

duced. The following is one method of doing this. Keep-

ing the instrument clamped in the meridian, the star is
observed at a certain distance from the middle vertical
thread and the time noted. This will give its hour angle,
and if we denote this by t (in seconds of time) the reduc-
tion is obtained by the formula

i (15 ty sin 1" sin 2 d

This is to be added to the observed zenith distance of a

southern star, or subtracted from that of a northern one,
and, in either case, half of it is to be added to the latitude.
 The Zenith Telescope. 83

REFRACTION.

When the zenith distances are small the refraction

varies as the tangent of the zenith distance.

Let r=a tan z

r'=a tan z'

Then r r=a (tan z tans)

sin (z z')
=a :
T
cos z cos z
a sin
= {z
.

z')-
,
r
1'
, nearly
J
cos 3 z
a may be taken as 57". 7, and the difference of the
micrometer readings used for (z z')
THE PORTABLE TRANSIT INSTRUMENT AS A ZENITH
TELESCOPE.
If the portable transit telescope has a micrometer
added to it, and the level of the finder circle is made

sufficiently delicate, it may be used as a zenith telescope,

reversing the instrument in its Ys between the observa-

tions.

Note The
catalogues give the mean places of the stars.
The apparent places are those which have to be used, and
must therefore be determined.
 CHAPTER IX.

ADDITIONAL METHODS OF FINDING THE LATITUDE.

TO FIND THE LATITUDE BY A SINGLE ALTITUDE TAKEN
AT A KNOWN TIME.

Here we have in the triangle PZ S the hour

angle P, the side Z S (go the objects alti-

tude), and P S the polar distance. From these
data we have P Z. From S draw S M
to find

perpendicular to P Z produced. Let d be the

declination, <p
the latitude, and a the altitude.
In the triangle PM S we have :

cos P=tan P M cot P S=tan P M tan d

M Z=P M P Z=P M+<p 90 cos ZS
Also, si***e '* ^=~~r^fs IcosZM
cos P M cos Z M cos P S cos Z S
: : : :

or cos P M sin (P M + ^)::sin d

: sin a :

Therefore
/ ** \
sm a cos P M
M+^)=
.

sin (P (2)
sin d

Equation (i) gives P M and (2) gives P M-f <p

In this method, if the star is observed when far from the
meridian a small error in the hour an^le produces a large
error in the computed value of the latitude. The altitude
should therefore be taken when the object is near the meri-
dian.
 Latitude by A Uitude of Pole Stay. 85

TO FIND THE LATITUDE BY OBSERVATIONS OF THE POLE

STAR OUT OF THE MERIDIAN.
If p be the polar distance of the pole star
in circular measure p 3 is a very small
quantity.

Let P be the pole, Z the zenith, and S

the star at an hour angle h or S P Z. Draw
S N at right angles to P Z and take Z M
equal to Z S. Let P N be denoted by x,
M N by y, S P by p, the star's altitude by
a, and the latitude by /. Then
Fig. 25.
P Z=Z M + M F=Z S-f-P N N M
or 90 l=go a+x y
;. l=a x-\-y
We have to find x and y.
(1) From the right-angled triangle S P N we have
cos S P N=tan P N cot P S
/. tan *=tan p cos h
or, approximately, x=p cos h
(2) Denoting S N by q we have from the same triangle
Sin S N=sin S P sin S P N
or sin ^=sin p sin h
:. approximately, q=p sin h.

(3) In the right-angled triangle S N Z we have

cos Z S=cos S N cos Z N
.*. sin tf=cos q sin (a+y)
.
, K sin#
or sin ya+y)-
cos #
or approximately
sin a
sin a+4 cos a=-
2
=sin a (i+J q )

y cos q a^ 2
sin a
or y == i Q z tan a
=i p* sin 3 h tan a
86 Circum -Meridian Altitudes.

Hence, in circular measure

l=a 2 2
p cos h-\-^ p sin h tan a
or in sexagesimal measure
l=a p cos h + \ p 2 sin i" sin 2 h tan a

This is the method given in the explanations at the end

of the Nautical Almanac. To find the latitude we have
only to take an altitude of Polaris, note the time (which
will give us the sidereal time), and apply certain correc-
tions as directed in the Almanac.

FINDING THE LATITUDE BY CIRCUM-MERIDIAN ALTITUDES.

When the latitude has been found by a single meridian
altitude the result only approximately true. It may,
is

however, be obtained with great exactness by taking a

number of altitudes of the sun or a star when within about
a quarter of an hour of the meridian on either side of it.
The altitudes may be taken with the sextant, reflecting
circle, or theodolite, and the observations should follow
each other quickly, and at about equal intervals of time.
The watch error must be exactly known, and the time
of each altitude noted. The mean of the altitudes is

taken, but the hour angle for each must be obtained

separately. In the case of the sun this is done by cor-
recting the observed times for watch error and subtracting
them from the mean time of apparent noon. If a star is
used the mean time corresponding to its R. A. will, of
course, give the hour angles
The formula is

Latitude=go* a d x"

Where a is the mean

of the altitudes, d the declination
of the object (negative if south), and x" a quantity equal to

2 sin 8
2
:
Xcos.approx. lat. x cos. dec'n. x sec. alt. ; h being

the hour angle.

 Circum-Meridian Altitudes. 87

To prove the formula

l=go ad x", when the latitude is approximately
known,
h
2 sm s* .

and x
, m
= :
2
w x
cos / cos d
sin 1 cos a
Let P be the pole, Z the zenith, and S the
sun or star near the meridian.
Let a be the star's altitude, h its hour angle,
and d its declination.
Let a + x be the star's meridian altitude.
Then a-\-XqI d=go
/

We have now to determine the small quantity x.

Now, sin PZ sin PS cos ZPS=cos ZS cos PZ <W^ 6 -

cos PS or cos / cos d cos /t=sin a sin / sin d

cos cos d cos h) = sin a + cos
:. / (1
= sin tf+sin (a
(/

+ x)
d)

j - h x x
.'. 2 cos
1
I cos d *
sin-1 =2 sin cos (a
,
-\
.

)
J
2 2 2

Therefore, approximately
.
a h
2 sin*
2 cos I cos d
# : -7F-X
sin 1 cos a
^ is, of course, negative if south.
h
2 sin's
The value of the expression _?
(known as the
sin 1

"reduction to the meridian") is found for each hour

angle from a table, and the mean of all the values taken
to calculate x" with.

Example On
a certain date, at a place the latitude of
which had been approximately ascertained to be 29" 52' o",
the mean of ten altitudes of the sun's lower limb, ob-
served with a powerful theodolite, was 39' 59' 20". This,
88 Circum-Meridian Altitudes.

when corrected for refraction, parallax, and semi-diameter,

gave 40 14' 31". 55 as the true mean altitude of the sun's
centre.The sun's declination was 19 53' 45".8 south.
The mean of the values of the reduction for the observed
hour angles, as taken from the table, was 16". 26, and the
calculated value of x was 17". 36.
90 o' o"
Altitude 40 14 31.55

49 45 28.45
Declination... 19 53 45.80

29 51 42.65

Latitude=29 51 25.29
Strictly speaking, a further correction ought to be made
for the change in the sun's declination during the obser-
vations.
In the case of a star we must add 0.0023715 to the log.
of x" to correct the hour angle for the difference between
the sidereal and mean time intervals; for the star moves
faster than the sun, and therefore gives a larger hour

angle for the same time.

Additional accuracy is obtained by taking half the ob-
servations east of the meridian, and half west of it, the
intervals of time between the successive observations being
made as nearly equal as possible. The hour angle
changes itssign after the meridian passage of the object.
 CHAPTER X.

INTERPOL A TION.

METHODS OF FINDING THE LONGITUDE.

In taking out variable quantities from the Nautical

Almanac it is necessary to interpolate for the local time

and longitude of the place of observation, since the data

given are for Greenwich time.
If the rate of change of the variable quantities is itself
variable we must allow for it if we wish to obtain a very
accurate result.

As an instance: we wish to find with great exactness the

sun's declination at apparent noon on the 2d January of
a certain year, at a place in longitude 6o or 4I1. west.

For Greenwich mean noon we find in the Almanac

Date. Sun's declination. Variation in 1 hour.
2nd January 22 57' i6".2 i3".2i
"
3rd 22 5i'45"4 i4"-35

Now, apparent noon at the place it will be 4 p.m.

at

apparent time at Greenwich, and^take the variation at

2 p.m. Greenwich as the average variation for those 4 hours.

This variation is i3'.305, which multiplied by 4 gives

go Interpolation.

53".22 to be subtracted from the de- 14-35

clination of 2d January 13.21
22 57' i6".2
53".22 1. 14
12)

22 56 22.98=required dec'n. .095

13.21

to
1 1 r\ f Variation
0'0J at2p.M

53-22

INTERPOLATION BY SECOND DIFFERENCES.

The differences between the successive values of the

quantities given in the Nautical Almanac as functions of
the time are called the first differences ; the differences be-
tween these successive differences are called second

differences; the differences of the second differences are

third differences, and so on. In simple interpolation we
assume the function to vary uniformly that is, that the ;

first difference is constant, and therefore that there is no

second difference. If this is not the case simple interpo-

lation will give an incorrect result, and we must resort to

interpolation by second differences, in which we take into

account the variation in the first difference, but assume
its variation to be constant and that there is no third

difference.

The formula employed is

f(a-{-k)=f(a) + Ak + Bk 2

where A is half the sum of two consecutive first differ-

ences and Bis half their difference. It is thus derived:

We have by Taylor's Theorem

/(.v + /0=/(-v) + A// + M/z 2 ! &C. (A)

and if h is small compared with x the successive terms of

the series grow rapidly 1'

 Interpolation. 9i

Suppose a 1, a, and a-\-i to be three successive argu-

ments of a table constructed from / (x) in which it is as-
sumed that a is many times greater than 1. Then, from
the table we know_/ (a i),/(a), and/(a + i), and there-
fore we know the differences f (a) f{a 1), and f(a-\-i)

f(a) which
,
we may designate by A and A' respectively.
Knowing that third differences can be neglected we can
obtain the value of/ (a + k), where k is less than 1, as
follows :

Fromwe have, if we
(A) neglect higher powers,
f (ai)=f (a)-A + B (1)

/(*)=/(*) (2)

/(a + i)=/(a)+A + B (3)

f (a+k)=f (a) +Ak + Bk* (4)

Subtracting equation (1) from (2) we get

A=A B
and subtracting (2) from (3) A'=A-f-B
/. A=4(A'+A)
B = i(A A)
.'.
substituting in (4)
/ (* + *)=/()+ (A'+ A)k + i (A' A) k*
The signs of A and B must be carefully noted. If the
functions are decreasing the first differences are negative,
and if the first differences are decreasing the second differ-
ences are negative.
The method can be better understood from an example
or two.
Ex. 1. Given the logs, of 365, 366, and 367 to 7 places
of decimals to determine log. 366.4.

Numbers.
92 Interpolation.

Here k is y4^, A =11866, and B= 16.

563481 1 11866^0
4746 4

3639557 47464
3
-x(A) 2
-3, nearly.
3639554
The tables give the log. as 3639555.

If the second difference had been neglected i.e., if we

had worked by simple interpolation, the result would have

been 563955 1 -

Ex. 2. Given the log. cosines of 89 32', 89 33', and

;

89 34, to find log. cos. 89 33 15".

 Interpolation. 93

To correct for second differences we have now only to

find the difference of R. A. for one minute at the middle
instant of the interval T' T. Call this A', and we shall
have

T'-T= =z
A
T and T' are in minutes.

INTERPOLATION BY DIFFERENCES OF ANY ORDER.

If it is required to find the intermediate values of a
function with greater exactness than can be done by
interpolation by second differences we can use any num-
ber of differences.

Let T, T-fz>, T-j-2 w, &c, be the arguments.

<
F, F<, F", &c,
94 Interpolation.

Monti's R. A. 1st Diff. 2nd Diff. yl Diff. tfhDiff

Mar. 5, oh 2ih. 580128s .39

" 28m.
5, i2h 22 27 15 .43
"
6, oh 22 55 25 .50
"
6, i2h 23 23 3 .39
"
7. oh 23 50 15 63
" i2ho
7, 17 9 83
 Longitude by Moon-Culminations. 95

and the interval of time noted between the transits of the

bright limb of the moon and of certain stars, the right
ascension and declination of which are nearly the same
as that of the moon at the time. It is not
necessary to
know either Greenwich or local time, but the rate of the
watch should be taken into account. The instants of
transit are noted, and the interval of time between them
is reduced from mean to sidereal time.

In the Nautical Almanac

are given, for every day of the
year, the sidereal times of transit at Greenwich of the
moon and of certain suitable stars, called "moon-culminat-
ing" stars; also the rate of change per hour (at the time
of transit) of the moon's R. A. As the moon moves
rapidly through the stars from west to east, it is evident
that a station not en the meridian of Greenwich the
interval between the two transits will be different to that
at Greenwich; and, the moon's rate of motion per hour
being known, a simple proportion will (if the station is
near the meridian of Greenwich) give the difference of
time between the station and Greenwich, and thence the
longitude. If the station is far from the meridian of

Greenwich a correction will have to be made for the

change in the rate of change of the moon's R. A. The
rate of change at the time of transit is found from the
Nautical Almanac by interpolation by second differences,
and the mean of the rates of change at Greenwich and at
the station is taken as the rate for the whole interval of
time between the transits.

An example will best illustrate the method :

At Kingston, Canada, on the 24th February, 1882, the

and of the moon's bright limb
transits of the star v Tauri
were observed at 6h. om. 23s., and 6h. im. 9s. respec-
tively, mean time. Difference, 46 seconds, or 46s. 12 in
sidereal units.
96 Longitude by Moon-Culminations.

r -r 1 * \
v Tauri...4h.
+ iom. 16S.62
Greenwich Transits
] MoonI ^ ^
Difference in sidereal time= 11m. 19s. 18
Add interval at Kingston^ 46 .12

Total change of moon's R.A--= 12m 55.3=7255.3

By interpolation by second differences the variation of
the moon's R.A. per hour at Kingston at the time of
transit was found to be 142s .23
At Greenwich it was 142 .68

2)284 .91

Mean rate of variation 142.455

725 3
xih. = 5h.09i6
-3

142.45
5I1. 5m. 29s. 76 west longitude.
It should be noted that in this case the moon was west
of the star at transit at Greenwich and east of it at

Kingston, having passed it in the interval.

The following a specimen of the part of the Nautical

is

Almanac relating to moon-culminating stars.

 Moon-Culminating Stars, 97

en
g8 Longitude by Lunar Distances.

The moon moves amongst the stars from west to east

at the rate of about lg a day. Its angular distance from
the sun or certain stars may therefore be taken as an in-
dication of Greenwich mean time at any instant the
moon being made use of as a clock in the sky to
in fact
show Greenwich mean time at the instant of observation.
The local mean time being also supposed to be known,
we have the requisite data for determining the longitude
of a station.

In the Nautical Almanac are given for every 3d hour of

G.M.T. the angular distances of the apparent centre of the
moon from the sun, the larger planets, and certain stars,
as they would appear from the centre of the earth. When
a lunar distance has been observed it has to be reduced
to the centre of the earth by clearing it of the effects of
parallax and refraction, and the numb e rs in the Nautical
Almanac give the exact Greenwich mean time at which
the objects would have the same distance.

to be noted that, though the combined effect of

It is

parallax and refraction increases the apparent altitude of

the sun or a star, in the case of the moon, owing to its near-
ness to the earth, the parallax is greater than the refrac-
tion, and the altitude is lessened.

Three observations are required one of the lunar dis-

tance, one of the moon's altitude, and one of the other
object's altitude. The altitudes need not be observed
with the same care as the distance. The clock time of
the observations must also be noted. The sextant is the
instrument generally used. All the observations can be taken
by one observer, but it is better to have three or four. If
one of the objects is at a proper distance from the meri-
dian the local mean time can be inferred from its altitude.
If it is too near the meridian the watch error must be
found by an altitude taken either before or after the lunar
observation.
 Longitude by Lunar Distances. 99

Four or five sets of observations should be made and

written down in their proper order:
Time by watch. Alt. of star. Alt. of moon's lower limb. Diat. of moon's far limb
1st obs'n
nd "
3rd "
4th

4) Total*.

Mean
If there is only one observer it is best to take the ob-
servations in the following order, noting the time by a
watch. 1st, alt. of sun, star or planet; 2d, alt. of moon;
3d, any odd number of distances; 4th, alt. of moon; 5th,
alt. of sun, star, or planet. Take the mean of the dis-
tances and of their times. Then reduce the altitudes to
the mean of the times; i.e., form the proportion differ-
ence of times of altitudes cliff, of alts. diff. between
: : :

time of 1st alt. and mean of the times a fourth number :

which is to be added to or subtracted from 1st alt. ac-

cording as increasing or diminishing. This will give
it is

the altitudes reduced to the mean of the times, or cor-

responding to that mean.
The altitudes cf moon and star must be corrected as
usual, and the augmented semi-diameter of the moon
added to the distance to give the distance of its centre.
The lunar distance has then to be cleared of the effects
of parallax and refraction.

TO DETERMINE THE LUNAR DISTANCE CLEARED OF

PARALLAX AND REFRACTION.
Let Z be the observer's zenith, Zm
and Zs the vertical circles in which the
moon and star are situated at the instant
of observation. Let m and s be their
observed places, M and S their places
after correction for parallax and refrac-
tion then Zw, Zs, and ms are found by
:

observation, andZ M and ZS are obtained

by correcting the observations. The ob- Pig. 27.
loo Longitude by Lunar Distances.

ject of the calculation is to determine M S.

Now, as the angle Z is common to the triangles mZs

and MZS, we can find Z from the triangle mZs in which
all the sides are known. Next, in triangle MZS there
are known
Z, M Z
S, and the included angle Z, from
which M S can be found. M S is the cleared lunar dis-
tance. The numerical work of this process is tedious.
The cleared distance having been obtained we proceed
in accordance with the rules given in the N.A.
The Greenwich mean time corresponding to the cleared
distance can be found either by a simple proportion or by
proportional logs.
It admits of proof that if D is the moon's semi-diameter
as seen from the centre of the earth (given in N.A.), D'
its semi-diameter as seen by a spectator in whose zenith

it is, D" its semi-diameter as seen at a point where its alti-

tude is a, then
D" D=(D' D) sin a, very nearly.
For details of the methods of finding differences of
longitude by the transportation of chronometers, and by
the electric telegraph, vide Chauvenet or Loomis.
 CHAPTER XI

MISCELL A NEO US.

TO FIND THE AMPLITUDE AND HOUR ANGLE OF A
GIVEN HEAVENLY BODY WHEN ON THE HORIZON.

The amplitude is the

angle that the plane of the
vertical circle through an object makes with the plane
of the prime vertical.

N S E
Let W
be the north, south,
east, and west points of the horizon
respectively; P the pole and H the ;

heavenly body. Suppose H to be

between N and W. Join P H.

Here W
H is the amplitude (a),
and in the triangle H P N we have
NP the latitude (<p),
HP the object's

polar distance (90 ),

and H N P a right angle. Also, if
t is the hour
angle, H P N==i8o t, and N H=goa.
Hence:
sin a= sec <p
sin d\
cos /= tan <p
tan d)

From the second of these equations we can calculate

the time at which the heavenly body rises and sets.
102 Parallax.

TO FIND THE EQUATORIAL HORIZONTAL PARALLAX OF A

HEAVENLY BODY AT A GIVEN DISTANCE FROM THE
CENTRE OF THE EARTH.
Referring to the figure in the next article, if A is the
observer's position H' will be the apparent position of the
heavenly body, and if C be the centre of the earth the
equatorial horizontal parallax will be the angle H'. Desig-
nating A C by r, A H' or C H
by d, and the parallax by p,
we have
sin p=-r
TO FIND THE PARALLAX IN ALTITUDE, THE EARTH BEING
REGARDED AS A SPHERE.
In Fig. 29 A is the observ-
er's position, Z
the zenith,
CH the rational horizon,
A H' the sensible horizon, and
S the heavenly body. Let p
be the horizontal parallax
(H'), /the parallax in altitude
(S), h the altitude (S A H'),
and d the distance of the tig 29.

heavenly body (S C). From the trian jle SAC we have

sin S sin S _AC
sin ZAS sin S AC S C

or sin^' = cos h sin p

The angles p and p' being (except in the case of the
moon) very small, we may substitute them for their sines,
and the equation becomes
'

p p cos h
STAR CATALOGUES.
If we want to find the position of a star not included

amongst the small number (197) given in the Nautical

Almanac we must refer to a star catalogue. In these
 Star Catalagues. 103

catalogues the stars are arranged in the order of their

right ascensions, with the data for finding their apparent
right ascensions and declinations at any given date. These
co-ordinates are always changing. 1st. by precession, nuta-
tion, and aberration, which cause only apparent changes of
position; 2ndly, by the proper motions of the stars them-
selves amongst each other. In the catalogues the stars
mean equator and
are referred to a a mean equinox at
some assumed epoch. The place of a star so referred is

called itsmean place at that time that of a star referred

;

to the true equator and true equinox its true place and ;

that in which the star appears to the observer in motion

its apparent place. The mean place at any time can be
found from that of the catalogue by applying the
precession and the proper motion for the time that has
elapsed since the epoch of the catalogue the true place will
;

then be found by correcting the mean place for nutation ;

and, lastly, the apparent place is found by correcting the
true place for aberration.
The most noteworthy star catalogues are the British
Association Catalogue (B. A. C.) containing 8,377 stars,
the Greenwich catalogues, Lalande's, containing nearly
50,000, Struve's, Argelander's, &c., &c.
DIFFERENTIAL VARIATIONS OF CO-ORDINATES.
It is often necessary in practical astronomy to deter-
mine what effect given variations of the data will produce
in the quantities computed from them. If the variations
are very small the simpler differential equations may be
used. The most useful differential formulae, as regards
spherical triangles, are deduced as follows :

We have the fundamental equations :

cos a=cos b cos c-fsin b sin c cos A \

sin a cos B = cos b sin c sin b cos c cos A

sin a sin B = sin b sin A (1)
sin a cos C = sin b cos c cos b sin c cos A
sin a sin C=sin c sin A
104 Differential Variations.

Differentiating the first equation of this group and

changing signs, we have
sin a i#=sin b cosci6-fcos b sin c dc cos b sin c cos A db
sin b cos c cos A dc + s'm b sin c sin A d A
= (sin b cos c cos b sin cos A) db +
c

(cos 6 sin c sin b cos cos A) dc-f-sin 6 sin c sin A i

c A
= sin a cos C db-\- sin # cos B rfc-hsin 6 sin c sin A da
ci ri A
or <fo=cos C ^6 + cos B dc-fsin 6 sin c-r- -d A
sin a
= cos C ^/c+sin 6 sin C d A
d6 + cos B
or da cos cos B dc=sin b sin
C db CiA ]
Similarly we obtain . ,

cos C da-\-db cos A dc s'm c s'm A d B = \

^ '

cos B da cos A db-\-dc=sin a sin B d C J

From these, by eliminating da, we obtain :

sinC db cos a sin B ^c=sin b cos C d A + sin a d B) , .

cosflsinC db-\-s\n B dc=sm afcos B d A-f sin a d CJ ^>

and by eliminating db from these :

sin a sin B ^c=-cos b d A-f-cos a d B-\-d C (4)

If we eliminate d A from (3) we get
cos b sin C db cos c sin B ic=sin c cos B d B
sin 6 cos C d C
and, by dividing this equation by sin b sin C or its

equivalent sin c sin B, we have

cot b db cot c ic=cot B d B cot C d C (5)

As an example, take the astronomical triangle PZS,

and put
A Z a=go d
B =* b=C
C = c=90 <p

Then the first equations of (2) and (3) give

d 8 cos q d d Z-f cos t d <p\
sin q sin
,^
cos Sdt= sin </ d d Z-fsin d sin d <p\ ^ '
cos q sin
which determine the errors d S and d t in the values of S
and /
computed according to the formulae
 Differential Variations. 105

sin &=sin <p

cos + cos <p
sin cos Z)
cos<5 cos =-cos cos sin sin cos
(p ^> Zf (7)
cos d sin *= Z) sin sin

(which are derived directly from the fundamental equa-

tions (1) ), when ^Z,and $,are affected by the small errors d,
d Z, and d <p respectively.
In a similar manner we obtain
d = cos q d S + sin q cos d d cos Zd<p \(o\
sin ^Z sin # d d + cos q cos d d t cos sinZ^^J
^ '

which determine the errors d and d Z in the values

of and Z computed by the formulce (derived as above)
cos =sin <p
sin + cos <p
cos d cos t\

sin J cos Z=cos <p sin sin <p

cos # cos $1 (9)
sin sin Z =cos d sin )

when d, t, and ^ are affected by the small errors d d,

d t, and d <p, respectively.
(It seems almost superfluous to point out that in these
formulce the latitude, d the star's declination, q the
tp
is

angle S, or parallactic angle, the star's zenith distance,

and t the star's hour angle.)
TO FIND THE CORRECTION FOR SMALL INEQUALITIES IN
THE ALTITUDES WHEN FINDING THE TIME BY
EQUAL ALTITUDES OF A FIXED STAR.
If from a change in the condition of the atmosphere
the refraction is different at the two observations, equal
apparent altitudes will not give equal true altitudes. To
find the change At hour angle t produced by a
in the

change A a in the altitude we have only to differentiate

a
the equation.
Sin =sin <p
sin (5+ cos <p
cos d cos t

Regarding <p
and d as constant : whence
Cos aAa= cos <p
cos d sin t
15 At
where A is in seconds of arc, and At in seconds of
time.
If the altitude at the west observation is the greater by
io6 Effect of Errors.

Aa the hour angle is increased by A*, and the middle

time is increased by ,
which is therefore the correction
2
for the difference of altitudes. From the above equation
its value is

A cos a
30 cos <p
cos S sin t

If A is the azimuth of the object, we have

cos d sin
Sin
_.
A
.

= cos a
t

and the formula may be written

30 cos <p
sin A
which will be least when the denominator is
greatest ;

that is, when A=90

or 270, The star is therefore best
observed on or near the prime vertical. Low altitudes
are, however, &esf owing to uncertainty in the refraction.
If the star's declinat'on is nearly equal to the latitude
the interval between tha observations will be short, which
is an advantage, as the instrument will be less liable to

change.
EFFECT OF ERRORS IN THE DATA UPON THE TIME COM-
PUTED FROM AN ALTITUDE.
We have, from the first differential equation (8), multi-
plying dt by 15 to reduce it to seconds of arc,

15 sin q cos d dt = d J-f-cos Z rty + cos q d 6

s

If the zenith distance above ^Ts erroneous we have

d<p=o, and d #=0, and

15 dt = ~. 2
3,= cos
^
L
rj
sin q cos 6 tp
sin
from which it follows that a given error in the: altitude will
have the least effect upon the time when the object is on
the prime vertical. Also, that these observations give
the most accurate results when the place is on the
equator, and the least accurate when at the poles.
 Effect of Errors. 107

By putting d C=o, dd=o, and sin q cos =cos <p

sin Z,
we have

15*- cos tan Z

<p

by which we see that an error in the latitude also pro-

duces the least effect when the star is on the prime verti-
cal, or the observer on the equator. In the former case
tan Z is infinite ; therefore, if the latitude is uncertain,
we can still get good results by observing stars near the
prime vertical.

If ^=0 and d<p=o we have

dd
I 5 dt
,.
= cos jr-tan
q
Hence an error in the star's declination produces the
least effect when the star is on the prime vertical (since
tan q is a maximum when sin Z=i), and that, of different
stars, thosenear the equator are the best to observe.
In high latitudes
it will often be necessary, in order to

avoid low altitudes, to observe stars at a distance from

the prime vertical. In this case small errors in the data
will affect the clock correction. But if the star is ob-
served on successive days on the same side of the meri-
dian at about the same azimuth, the clock's rate will be
accurately obtained, though its actual error will be un-
certain.
If the same star is observed both east and west of the
meridian, and at the same distance from it, constant
errors d<p, dd, and d, will give the same value of dt, but
with opposite signs. Hence one clock correction will be
too large, and the other too small, and by the same
amount, and their mean will be the true clock correction
at the time of the star's meridian transit.

EFFECT OF ERRORS OF ZENITH DISTANCE, DECLINATION,

AND TIME, UPON THE LATITUDE FOUND BY
CIRCUM-MERIDIAN ALTITUDES.
The formula for finding the meridian zenith distance

f' from a circum-meridian zenith distance is

io8 The probable error.

2 sin 2
,
where A
.

= cos <p
1
cos o
-? and
.

m = rir-
v^
sin f sin I

Differentiating this, and regarding A as constant, we

have, since d<p=d + dd

dadr+dd
T * -A^i 5 dt
sin i"
J

The
errors d and dd affect the resulting latitude by
their whole amount. The coefficient of dt has opposite
signs for east and west hour angles therefore, if obser- ;

vations are taken of a number

of pairs of equal altitudes
east and west of the meridian, a small constant error in
the hour angles (or clock correction) will be eliminated
in the mean. This result is practically attained by
taking the same number of observations at each side of
the meridian, and at nearly equal intervals of time.

An error in the assumed latitude which affects A is

eliminated by repeating the computation with the latitude

found by the first one.

THE PROBABLE ERROR.

To what is meant by the term "proba-
give an idea of
ble error," we
suppose a rifleman to have fired a
will

large number of bullets at a target at the same range and

with equal care in aiming, and that, on examining the
target, it is found that half of them have struck within
three feet of the centre, and half outside that radius ;

then it may be assumed, a chances of any

priori, that the
one shot hitting within 3 feet of the centre are even in
other words, that it is an even chance whether or not the
bullet will strike within that distance or not. And this
distance may be taken as the probable error of any one
shot.

Now, if we make
a series of independent but equally
careful measurements of a given quantity, such as an
 The probable error. 109

angle or a base line, they will all differ more or less, the
closeness of the agreement depending on the instruments
employed and the care exercised; and the problem is to
decide what value is to be taken as the most likely to be
the correct one in other words to have the smallest
probable error.
If Aj, a 2 , 3 measurements, n their
,&c, are the different
a x + a 2 + &c. + a Q -

number, and m their mean, then m= l

, , , .
;

n
and it follows as an arithmetical consequence that the
algebraical sum of the quantities (m a x ), (m a 9 ), &c,

(m a n ) will be equal to zero. These quantities are

called the "residuals.'" Another property of the mean is
that the sum of the squares of the residuals, (m a x ) 2 (in a 2 ) 9 , ,

&c. is a minimum.
Now it admits of proof that the mean is that value, de-

rived from the various measurements, which is likely to

be nearest the trutn. The value of the probable error of
the mean is

V(m at)3 +(m a 2 )* &c, + (m a n )*

x
^~ 0.674489
(I) #
And the probable error of any one measurement is the
probable error of the mean multiplied by |/~^
It must be borne in mind that by the probable error
being taken as so much is meant that it is an even chance
that the value taken is within that much of the truth
without regard to sign. Thus, mean of a if / be the
number of measurements of a base line, and 1 foot its

probable error, it is an even chance that its real value

lies between / 1 and l+i.

Instead of using the probable error of a result we often

employ what is called its weight; a function which indi-
cates the relative value to be assigned to the results as
regards precision.
no The probable error.

The formula for the weight is

n2
2 (2)
a (m
{
aj ) + (m a 2 ) 2 + &c. +, (in an )
2

6
\ Probable error=^Zgl
4/ weight

So that the weight varies inversely as the square of the

probable error. From the property of the sum of the squares
of the residuals being a minimum in the case of the mean,
this method is often called the "method of least squares."

As a simple example of the calculation of the probable

error we take a side of a triangle forming part of a
will

triangulation carried out near Kingston in 1881-82. Four

independent measurements were made to ascertain its

length,and the results were:

I
 Part II

GEODESY.

CHAPTER I.

THE FIGURE OF THE EARTH.

Geodesy is a word of Greek derivation, and signifies

"division of the earth." Broadly speaking, it comprises
allsurveying operations of such magnitude that the fig-
ure of the earth has to be taken into consideration.

The earth is an oblate spheroid that is, the figure

formed by the revolution of an ellipse round its
minor axis the polar axis being shorter than the
equatorial by about 26*88 miles. This has been
proved in two ways. Firstly, by pendulum experiments,
which show that the force of gravity increases from the
equator towards the poles secondly, by actual measure-
;
112 Figure of the earth.

ments of portions of meridianal arcs. A little considera-

tion will show that if the curvature of a meridianal arc is
elliptical, and therefore decreasing towards the poles, the

length on the earth's surface of a degree of latitude must

be greater in high than in low latitudes. That is, if A
and B are two points on a meridian near the equator,
but differing by a certain amount in astronomical latitude,
and C and D two points on a meridian in a high latitude
and also differing by the same amount, then if the dis-
tances A B and C D are measured on the ground, A B
will be found to be less than C D. This has actually
been done at various parts of the earth's surface Lap-
land, Peru, France, Russia, (where an arc of over 25was
measured,) India, Algiers, South Africa, &c. The method
adopted is to measure a base very accurately, and
from it to connect by means of a chain of triangulation
two distant stations which are as nearly as possible on
the same meridian. This being done we can calculate
the actual distance from one of the stations to the point
where a perpendicular drawn to the meridian of that sta-
tion from the other station meets the meridian. The
latitudes of the two stations are found by very careful
astronomical observations, and their difference, taken in
connection with the calculated distance on the meridian,
gives the curvature of the arc, since the radius of curva-
ture is the measured distance divided by the difference of
latitude in circular measure. There is, however, one
source of error in determinations of this kind. In finding
the latitudes of stations we
are in general dependent on
the direction of the plumb line; and should there, as
often happens, be a local abnormal deviation of the
latterfrom the true perpendicular, the resulting latitude
will be erroneous. This was proved many years ago by
taking the latitudes of two stations on opposite sides of a
mountain in Perthshire, and measuring the true horizon-
tal distance between them, when it was found that the
 Figure of the Earth. 113

difference of the latitudes as given by the astronomical

observations was by no means the true one.

The experiment was repeated in 1855 at Arthur's seat

near Edinburgh. Three stations, A,B,and C, were taken ;

B on the summit of the hill, and the other two at opposite

sides of it. The astronomical differences of latitude of
the stations, as obtained by a long series of observations,
were as follows :

Betweeen A and B 25^.53

BandC 17"

while the actual differences, as found by triangulation,

were :

Between A and B 24^27

" B and C H"- 10
^

As a rule, the deviation seldom exceeds a few seconds

except the neighbourhood of great mountain masses,
in
as at the foot of the Himalayas, where it is as much as
30".

Where there is considerable deviation in level countries

it no doubt caused by neighbouring portions of the
is

eartn's crust being either denser or lighter than the aver-

age. The actual amount of deviation at a station may
be ascertained as follows. A series of other stations
symmetrically grouped round it are chosen, and the latitude
and longitude of each obtained by astronomical observa-
tions. The actual distance and azimuth of the central
station from each of the others being known by triangu-
lation, itslatitude and longitude may be calculated from
that of each of them separately, and the mean of
the results may be taken as nearly the truth, since
the errors caused by plumb line deviations at the various
stations will probably counteract each other. This mean
latitude and longitude of the central station being com-
pared with the latitude and longitude as obtained by
ii 4 Figure of the Earth.

astronomical observations will give the deviation of the

plumb line.

If a and b are the semi-major and semi-minor axes of an

ellipse, the distance of the centre from either focus is

|/# 2, and this quantity divided by a is called the

"eccentricity." This is generally written e. The quantity

is called the "compression" or "ellipticity," and is

denoted by c. The latest calculations make the com-

pression of the earth about -jfe the ratio of the semi-
axes being believed to be 292 to 293. The true measure of
the compression is the difference of the semi-axes divided
by the mean radius of curvature of the spheroid. The
equator has also been found to be elliptical, its major
axis being about 500 yards longer than its minor axis.

should be noted that the expression e has different

It

meanings in different books. English writers occasion-

ally employ it for the compression or ellipticity, while in
American books it is used in the same sense as here,
namely, for the eccentricity. Even chapters of
in different

the same work the letter e is often used both for the
compression and the eccentricity.
The accompanying
figure represents a
section of the earth.
PP' is the polar axis,
QE an equatorial di-
ameter, C the centre,
F a focus of the

ellipse, A
a point on
the surface, A T a
tangent at A, and
Z A O perpendicular Fi8- 3-

to A T. Z' is the geocentric zenith, and Z' C E' is its

declination. The latter is called the geocentric or reduced lati-

 Figure of the Earth. 115

tude of A. Z O' E' is the geographical or astronomical latitude

Z A Z' or C A O is called the reduction of the latitude.
It is evident that the geocentric is always less than the
geographical latitude.
Let C K=a. C P=b. Let c be the compression and e

the eccentricity.
ab b
a a
n6 Figure of the Earth.

in which <7=T-r-
/+*
Applying this to the development of (2) we find, after
dividing by sin 1" to reduce the terms of the series to
seconds, and putting x=tp' y=><p. f

r
tp r
<p
= ^2 - S in 2 r
tp --. sin 4 tp
^ r &c. (3)
VJ
sin 1 2 sin 1"

in which q= r+ i 2
p 1 e 2 -f 1 2 <;

The known value of e gives q, and thence ^ tp'

for

any given value of tp.

N.B. q negative, and #

is 2 is very small compared
with it ; therefore tp tp is positive.

In some books on geodesy th? expression "reduc-t+en df

tke latitude" is applied to the angle A' C E, where A' is

the point in which B A proiuc3d meets the circle de-

scribed with centre C and radius C E. Let this angle be tp".

_, tan tp'
z-=
B A b
Then T^-r ,=
-

tan tp" a BA <

/
r
by the properties of the ellipse. And since tan p' *-tar tp

,
we have
tan tp
T
tan tp
= a*
r-tan r
.

tp
,
~-
.
a
tan c>'
r
,
a
b2 b b
tan y tan
,
_ tp"

tan ^" tan ^'

TO FIND THE RADIUS OF THE TERRESTRIAL SPHEROID FOR

A GIVEN LATITUDE.
Let p (or A C) be the radius for latitude tp.

We have, p |/ x a + y
To express a: and j/ in terms of y?,
we have, substituting
b2
1 g* for
j
in the equation to the ellipse and its differ-

ential equation,
 Figure of the E arth .
117

2
1 e

X=(i_s)tai
whence, by elimination, we find
cos
#= y>

ri-r sin-1 <p

2
(1 e ) a sin <p
2
Vi e sin 2 <p

4
/i 2 e* sin 2 <p-t-^ sin 2 a\J- (4)
and hence, P=a[ _ .
.
I
2
sin 3
1 ]
V 1 <p J

TO FIND THE LENGTH OF THE GREAT NORMAL, A O, FOR

A GIVEN LATITUDE.
From the figure we have
Great normal= -
cos <p

a
"
|/i_g3 si n 3 p (5)

In future the great normal will be spoken of simply as

the "normal."

Note. The length of a second of longitude in latitude

<p
will be x sin 1", or
a cos sin 1"
T
<p
=1 JV'cos y> sin, J.,
2
Vi e* sin <p

TO FIND THE RADIUS OF CURVATURE OF THE TERRESTRIAL

MERIDIAN FOR A GIVEN LATITUDE.
Denote this radius by R.
We have, from the Differential Calculus,

r . i + m '

dx*
n8 Figure of the Earth.

From the equation to the ellipse we have

dy _ b* x
dx a* y
d^y _" J^
dx 2 -
a 2 y*

whence R-^+y*)*
2
), we find, by substitut-
Observing that 6 a 2 (i 2

ing the values of x and y in terms of <p (page 115.)

r *( x ~ g2 )
m yi-'/ (6)
C6 "
(1 <?
2 sin 2 ^)f
This last equation gives the length of a second of lati-

tude at a given latitude, since it is equal to R sin 1"

The following formula is sometimes used for the radius

of curvature of the meridian,

D a -\-b
K ,
f
,
(a
,

0)
x
cos 2 9?

It also admits of proof that the normal at any point is

the radius of curvature of a section of the earth's surface
through the normal and at right angles to the meridian.
From equations (5) and (6) we see that the normal at
any point always greater than the radius of curvature
is

of the meridian at that point.

If the earth were a sphere the shortest line on its sur-

face between any two points A and B (otherwise called
the .geodesic line) would be an arc of a great circle, and
the azimuth of A at B would differ from that of B at A
by 180. ^uT'o^frie^uHkc^o'f a sprteroicf The geodesic
line is, except when both points are on the equator or on
the same meridian, a curve of double curvature. The two
azimuths, also, will not, except An certain cases, differ
<>
from each other by exactly *8e-. Thereason oTthis is
that the vertical plane at A passing through B will not
 Figure of the Earth. 119

coincide with the vertical plane at B passing through A.

These two planes will, of course, intersect at A and B,
but their intersections with the surface of the spheroid
will be different curves, and will enclose between them a
space. In addition to these two lines and the geodesic
line there will also be what is known as the line of align-
ment of the two points that is the line on every point of
which the line of sight of the telescope of a theodolite in

perfect adjustment and truly levelled would, when

directed on one station, intersect the other on the tele-
scope being turned over.
 CHAPTER II.

GEODETICAL OPERATIONS.

The methods adopted in the old world for mapping

large tracts of country have been reversed in America.

Instead of starting from carefully measured bases, and
carrying out chains of triangulation connecting various
principal points in such a manner that the relative
positions of the latter with respect to each other may be
ascertained within a few inches, though several hundred
miles apart, the system pursued (if we except the U. S.
Coast Survey and some other triangulations) has been
to take certain meridians and parallels of latitude inter-
secting each other to trace and mark out these meridians
;

and parallels on the ground ;

to divide the figures
enclosed by them into blocks or "checks ;" and to further
subdivide the latter into townships, sections, and quarter
sections. Although the method of triangulation is incom-
parably the most accurate, the American plan has the
advantage of rapidity and cheapness. As the latter is
very simple, and is fully explained in the Canadian Gov-
ernment Manual of Survey, it will not be further touched
upon here.

At the commencement of a triangulation a piece of

tolerably level ground having been selected, a base line,
 Triangulation . 121

generally from 5 to 10 miles long, is measured with the

very greatest care, and from it either a network or a
chain of triangles is started. In the former case the
triangles are expanded as rapidly as possible till they are
large enough to cover the whole country with a network
of primary triangles. This is done by taking angles from
the extremities of the base to certain selected points or
"signals," (often on mountain tops), and calculating their
distances by trigonmetry. The instrument is then placed
at each of these new stations, and angles taken from
them to still more distant points, the calculated lines
being used as new base This process is repeated
lines.
and extended till the whole district
is covered by these

primary triangles_the sides of which should be as large as

possible.
Smaller, or secondary, triangles are formed within the
primary ones to fix the positionsof important points
which may serve as starting points for traverses, &c.
Tertiary triangles are sometimes formed within the secon-
dary ones.
The size of the
primary triangles varies according to
circumstances. Their sides are often from 30 to 60, or
even 100 miles, and in one instance as much as 170 miles.
The longest side in the British triangulation was III.
The sides of the secondary triangles are from about 5 to
20 miles, and those of the tertiary triangles five or less.

The larger triangles should be as nearly equilateral as

circumstances admit of. The reason for having them so
is that with this form small errors in the measurement
of their angles will have a minimum effect on the calcu-
lated lengths of the sides. Such triangles are called
"well-conditioned" ones.

The original base has to be reduced to the level of the

sea that is, the true distance between the points where
verticals through its ends intersect the sea level must be
122 Triangulation.

ascertained. The exact geographical position of one end,

and the azimuth of the other with respect to it, must of
course be known. The angles of all the principal triangles
must be measured with the greatest exactness that the
best instruments admit of, the lengths of the sides calcu-

lated by trigonometry, and their azimuths worked out.

The work (when carried on on a very large scale) is still
further complicated by the earth's surface being not a
sphere but a spheroid. The accuracy of the triangulation
is tested by what is called a "base of verification." That is,
a side of one of the small triangles is made to lie on
suitaole ground, where it can be actually measured. Its

length, as thus obtained, compared with that given by

calculation through the chain of triangles, shows what
reliance can be placed on the intermediate work.

As instances: The triangulation commenced at the

Lough Foyle base in the North of Ireland was carried

through a long chain of triangles to a base of verification
on Salisbury plain, and the actual measured length of the
latter to differ only 5 inches from the length as
was found
calculated through several hundred miles of triangulation.
An original base was measured at Fire Island, near New
York, and afterwards connected with a base of verifica-
tion on Kent Island in Ches'peake Bay. The actual
distance between them was 208 miles, and the distance
through the 32 intervening triangles 320. The difference
between the computed and measured lengths of the base
of verification was only 4 inches. In Algiers, two bases
about 10 kilometres long were connected by a chain of 58
triangles. Their calculated and measured distances
agreed within 16 inches.
If the country to be triangulated is very extensive
as, for instance, in the case of India instead of covering
it with a network of triangulation, it
may,, intersected in
the first place by chains of triangles, either single or double,
 Base Lines. 123

and bases measured at certain places^usually where these

chains meet. In India the chains run generally either
north and south or east and west, and form a great frame
or lattice work on which to found the further survey of
the country. A double chain of triangles forms, of
course, a series of quadrilateral figures, in each of which
both the diagonals, as well as the sides, may be calculated.

The following is a brief account of the measurements

of some celebrated base lines :

In 1736 a base line had to be measured in Lapland for

the purpose of finding the length of an arc of the meridian
by triangulation. A distance of about 9 miles was meas-
ured in mid winter on the frozen surface of the River
Tornea. By means of a standard toise brought from
France, a length of exactly 5 toises (about 32 feet) was
marked on the inside wall of a hut, and eight rods of pine,
terminated with metal studs for contact, cut to this exact
length. It had been previously ascertained that changes
of temperature had no apparent effect on their length.
The surveying party was divided into two, each taking
four rods, and two independent measurements of the base
were made, the results agreeing within four inches. The
time occupied was seven days. The rods were probably
placed end to end on the surface of the snow.

The same year a base 7.6 miles long was measured near

Quito in Peru, at an altitude of nearly 8000 feet. The

work occupied 29 days. Rods 20 feet long, terminated
at each end by copper plates for contact, were used.
The rods were laid horizontally, changes of level being
effected by a plummet suspended by a fine hair. The
rods were compared daily with a toise marked on an
iron bar which had been laid off from a standard toise
brought from Paris. This base was the commencement
of a chain of triangles for the measurement of a meri-
dianal arc. Three years later another base, 6.4 miles long,
124 Base Zincs.

was measured near the south end of this chain, and only
occupied ten days. The party was divided into two com-
panies which measured tne line in opposite directions.

The trigonometrical survey of Great Britain was com-

menced by the measurement of a base on Hounslow
Heath, which was chosen from the great evenness and
openness of the ground. Three deal rods, tipped with
bell metal and 20 feet long, were used at first. But it was
found that they were so affected by changes in the hu-
midity of the atmosphere that glass tubes of the same
length, of which the expansion for temperature had been
ascertained, were substituted, the temperatures of the
tubes being obtained by attached thermometers. The
length of the base when reduced to the sea level and 62*
Faht. was 9,134! yards. This distance was subsequently
re-measured with a steel chain 100 feet long, consisting of
40 links half an inch square in section. A second similar
chain was used as a standard of comparison. The chain
was laid in five deal coffers carried on trestles, and was
kept stretched by a weight of 28 pounds. The exact end
of each chain's length was marked by a slider on the top
of a post. The two measurements (glass tubes and
steel chains) agreed within two inches.

Two bases, each about j\ miles long, were subse-

quently measured in France one near Paris, the other at
Carcassonne in the south. Four rods were used. They
were composed of two strips of metal in contact (platinum
and copper), forming a metallic thermometer, carried en
a stout beam of wood. Each rod was supported on two
iron tripods fitted with levelling screws, and there was an

arrangement for measuring their inclination.

The Lough Foyle base was measured with Colby's
compensation bars; an arrangement in which the unequal
expansions and contractions of two parallel bars of differ-
ent metals (brass and iron), 10 feet long, are utilized to keep
 Base Lines. 125

two platinum points at an invariable distance from each

other. The bars were arranged in line on roller supports
in boxes laid on trestles, and the intervals between the
bars were measured by similar short compensating bars
six inches long, at each end of which was a microscope
into the focus of which the platinum points of the measur-
ing rods were bought by means of micrometer screws.
The line commenced at a platinum dot let into a stone

pillar,and the rods were kept in a true straight line

by a
transit theodolite. About 250 feet a day was measured
on an average; 400 feet of the line was across a river,
the boxes being laid on piles about 5 feet apart. Eight
miles were measured thus, and two more were subse-
quently added to the base by triangulation. The Salis-
bury Plain base was measured in the same way. Colby's
bars were subsequently used for ten bases in India, but
were not found to give very reliable results there.
An improvement on Colby's arrangement is the compen-
sating apparatus used in the United States coast survey.
It consists of a bar of brass and a bar of iron, a little less
than six metres long and parallel to each other. The bars
are joined together at one end, but free to move at the
other. Their cross-sections are so arranged that while
they have equal absorbing surfaces their masses are in-
versely as this specific heats, allowance being made for
their difference of conducting power. The brass bar is
the lowest, and is carried on rollers mounted in suspend-
ing stirups. The iron bar rests on small rollers fastened
to it which run on the brass bar.
126 Base Lines.

The annexed figure shows the arrangement at the two

ends, the hand part being the compensation end. It
left

will be seen that the lever of compensation (/) is pivoted

on the lower bar (a), a knife edge on its inner side abutting
on the end of the iron bar (b.) This lever terminates at
its upper end in a knife edge (c) in such a position that

whatever be the expansion or contraction of the bars it

always retains an invariable distance from their other
end. This knife edge presses against a collar in
the sliding rod (d),moving in a frame (/) fixed to the
iron bar, and is kept back by the spiral spring (s). The
rod tipped with an agate plane (p) for contact.
is The
vernier (v) serves to read off the difference of lengths of
the bars as a check.

At the other end where the bars are united a sliding

rod terminates in a blunt horizontal knife edge (g), its
inner edge abutting against a contact lever (h) pivoted at
(i). This lever, when pressed by the sliding rod, comes
in contact with the short tail of the level (k) ,
which is

mounted on trunnions and not balanced. For a certain

position of the sliding rod this bubble comes to the centre,
and this position gives the true length of the measuring
bar. Another use of the level is to ensure a constant
pressure at the points of contact, p and g. To the lever
and level is attached the arm of a sector which gives the
inclination of the bar.
 Base Lines. 127

The bars are enclosed in a spar-shaped double tin

tubular case, the air-chamber between the two cases pre-
venting rapid changes of temperature. The ends are
closed, the ends only of the sliding rods projecting. The
level, sector, and vernier, are read through glass doors.
The tubes are painted white and mounted on a pair of
trestles. Twoof these bars are used in measuring. They
are aligned by a transit.

On one base, seven miles long, measured with this ap-

paratus, the greatest supposable error was computed,
from re-measurement, to be less than six-tenths of an
inch. On another base, six and three quarter miles long,
the probable error was less than one-tenth of an inch,
and the greatest supposable error less than three-tenths.
This apparatus has been tested by measuring a base in
Georgia three times, twice in winter and once in summer,
at temperatures ranging from 18 to 107 Faht. The
discrepancies of the three measures with their respective
means were, in millimetres, 8.10, 0.32, and +8.41.
It has been found, however, that the apparatus is not
quite perfect, its true length depending on whether the
temperature is rising or falling.

The amount of accuracy to be aimed at in measuring

a base depends on the extent of the survey. For small
surveys it may be sufficient to measure the base two or
three times with a steel tape which is kept compared
with a standard. The tape should be stretched each
time to a constant tension by means of a spring balance.
It is a good plan to mark the end of each chain on a
small piece of plank, which is made to adhere to the
ground by means of pointed spikes on its under surface.

Pine rods, well seasoned, baked, boiled in drying oil,

painted and varnished, may be used. They should either
be levelled or have their angle of inclination read. If the
128 Base Lines.

ground is uneven they

be levelled on trestles with
may
sliding telescopic supports. The ends
of the rods should
be capped with metal, either wedge-shaped or hemi-
spherical in form, and either placed in actual contact, or
the spaces between them measured by graduated glass
wedges. If the end of one rod has to be placed on a dif-

ferent level to that of the next a fine plumb may be

line
used ;
or the rods may have fine lines marked at each
end of the unit of length, so that one rod may be made to
overlap the other with the two marks exactly correspond-
ing. This plan answers well on ice.

Before measuring an important base it is usual to make

a preliminary approximate measurement of the line, and
also to get an accurate section of it by levelling. Suit-
able points are selected for dividing it into sections, and
these points are accurately adjusted into line by means of
a transit at one end. It may happen, however, that it is

impracticable to have all the segments in a straight line,

in which case the angles they make with each other must,
of course, be exactly measured. Any deviation also from
a true horizontal line must be recorded in order that the
base may be reduced to the sea level. The ends of the
base, as well as of the sections, are generally marked by
microscopic dots on metallic plates let into massive
stones embedded in masonry, and are thus permanently re-
corded. The mark itself may be a minute cross on a piece
of brass, or a dot on the end of a platinum wire set
vertically in a piece of lead run into a hole in the stone.

If the rods used measuring the base expand and con-

in
tract with changes of temperature the latter must be re-
corded at regular intervals of time, as the rods are at
their true length only when at a certain standard tempera-
tuiv.

If the base, or any portion of it, is not level, its inclina-

 Base Lines. 129

tion must be measured for the purpose of reducing it to

its horizontal projection.

Let B be the length of an inclined portion, b the length

reduced to the horizontal, and 6 the angle of inclination.
Then b=B cos 6.

As 6 is generally a very small angle and need not be

known exactly, it is better to compute the excess of B
above b. If 6 is given in minutes we have
a
B 6 B (1 cos 6)=2 B sin 3
2
=4 B 6* sin 3 1'
2
=0.00000004231 6 B.
If the base is intersected by a ravine or creek which
cannot be conveniently measured across we may proceed
as follows :

Let AB CD be
the base, and BC
the interrupted
portion (Fig. 32).
Let AB^a, CD
= b, and BC^.r.

Take an exterior
station E and
measure the
angles AEB (a)
AEC (ft and AED Fig. 32.

(7-).
Then if <p is such an angle that

dab -sin 8 sin

r (r
v/ a)'
tan 2 <p= / 3
x .
,
(a b) sin a sin (7- p)
It may be proved that
a+b a b
x==
2 2 cos <p

The base is, of course, a + b+x.

130 Base Lines.

ground necessitates an angle C be-

If the nature of the
tween two portions of the base A C, C B, we can find
the direct distance The angle C (which
A B thus : is very
obtuse) is measured with great care. Let i8o C=0,
A C = b, C B = a, A B = c.
and
Then c 2 = a 2 +6 2 + 2 ab cos
and (if is not more than io)
62
cos =1 , nearly.

.\c 2
=a +& +2tf b a
2 2
b 6*
= (a + b)*ab6 2
,

("+ 6)
, ,v /
- J
a b 2
x

( j^w)
and
^(a+b)^^^-]
= &c
(
"+4-^w-+' -'l

,
, ab 6* sin 2 r
=a-\-b
2 (a+b)

=0 + 0.00000004231^ ?

being in minutes.
To reduce a measured base to the sea level we must
know the height of every portion of it in order to get its

mean height. Let I be the length of a rod, and/* its height;

/' its
projection on the sea level, and r the radius of the
earth.

or
/-^/{x-A), nearly.
If w be the number of rods in the base and n l=L;
then the length of the base reduced to the sea level will be

L -
J i /;
being the mean height of all the

rods.
 Base Lines. 131

The base thus reduced is a curve. To find the length

of its chord we should have to sujbtract a very minute

quantity, namely, the^ase^mvided by 24 times the mean

of the earth's radius.

MEASUREMENT OF BASES BY SOUND.

This is a rough method which has sometimes to be

adopted hydrographic surveys of extensive shoals which

in
have no points above water. It should, if possible, only
be adopted in calm dry weather. The velocity of sound
in air is1089.42 feet per second at 32 Faht. It isA un-

effected by the wind, the baromet^r^pressure, and the

hygrometic condition of the air. The observers are
posted at both ends of the base and are provided with
guns, watches, and thermometers. When the gun at one
end is fired the observer at the other notes the interval in
seconds and fractions between the flash and the report.
The guns are fired alternately from both ends at least
three times, a preparatory signal being given.

The value of the velocity of sound given above must be

corrected for the temperature (f) by multiplying it by the

quantity
Vi+{t 32)Xo-oo2o8.
Of course the distance is the corrected velocity multi-
plied by the mean of the observed intervals of time. The
errors of observation are always considerable, but are no

greater for long distances than for short ones.

ASTRONOMICAL BASE LINES.
In cases where no suitable ground for a measured base
is available two convenient stations may be selected as
the ends of an imaginary base line, and their latitude
and longitude, with the azimuth of one from the other,
ascertained by astronomical observations. shall We
then have the length and position of the base with more
or less accuracy, and a triangulation can be carried on
132 Base Lines.

from it. The base chosen should be as long as possible,

but not greater than one degree. None of the sides of
the triangles should be greater than the base. The azi-
muths of the sides being known, the positions of the
observed points can be plotted by co-ordinates.
If the zenith telescope and portable transit telescope
are used the latitude can be determined within io", the
longitude and azimuth within 30". With the sextant
these errors are at least doubled. Differences of longitude
may be determined by flashing signals.
 CHAPTER III,

TRIANGULATION.

Having discussed the measurement of base lines we

have now to consider the triangulation. It is evident
that the latter may be commenced without waiting to com-
plete the former. The first thing to be done is to select
the stations and to erect the necessary points to be ob-
served, or "signals" as they are called. In a hilly country
the mountain tops naturally offer the best stations, as being
conspicuous objects and affording the most distant views.
In this case the size of the triangles is only limited by the
distance at which the signals can be observed. Thus,
in the Ordnance Survey of Ireland the average length of
the sides of the primary triangles was 60 miles, while some
were as much as 100. In the triangulation which was
carried in i87g*across the Mediterranean between Spain
and Africa, by means of the electric light, signals were
observed at a distance of 170 miles.

In a flat country lofty signals have to be erected, not

only that they may be mutually visible, but in order that
the rays of light may not pass too close to the surface of
the earth, as they would be thereby too much affected by
refraction. In the U.S. Coast Survey six feet is consid-
ered the limit advisable. If h\ h", are the heights of two
134 Triangula Hon .

signals in feet and d their distance in miles, then, on a

flat country cr over water, they will not, under ordinary

circumstances, be visible to each other if d is greater than

$ ( \/h' + \ZK') The most difficult country of all in which

to carry out a triangulation is one that is flat and covered
with forest.

Formerly, conspicuous objects, such as the points of

church spires, were commonly used as signals; but of late
years this has not been done, because in all large triangles
it is necessary to measure all the three angles, and this

cannot well be done directly in the case of such objects.

The form of the signals varies much. Whatever kind be
used the centre of the theodolite must be placed exactly
under or over the centre of the station, and if a scaffold-
ing has to be employed the portion on which the instru-
ment is supported must be disconnected with that on
which the observers stand. One kind of signal is a ver-
tical pole with tripod supports, the pole being set up with
its summit exactly over the station. It may be sur-
mounted by two circular disks of iron at right angles to
each other. A piece of square boarding, painted white
with a vertical black stripe about four inches wide, can be
seen a long way off. Flags may be used, but are not al-
ways easy to see. A good form of signal is a hemisphere
of silvered copper with its axis vertical. This will reflect
the rays of the sun in whatever position the latter may be,
*
but a correction for 'phase" will be required, as the rays
will be reflected from different parts of the hemisphere

according to the time of day. The ordinary signal used

in the United States is a pole 10 to 25 feet high, sur-
mounted by a flag, and steadied by braces. With respect
to its diameter, the rule is that for triangles with sides
not exceeding live miles it should not be more than five
inches. If more than five miles, five to eight inches.
Various other forms of special signals are used in the U. S.
Coast Survey. Amongst others may be mentioned a
 Triangidation. 135

pyramid of four poles, with its upper portion boarded over

and terminating in a point, directly under which the theo-
dolite is placed. In England double scaffoldings as
high as 80 feet were used, the inner scaffolding carrying
the instrument and the outer one the observers. In
Russia a triangulation had to be carried on over an arc
of more than 500 miles across a flat swampy country
covered with impenetrable forests, and scaffoldings of as
much as 146 feet high had to be erected. On the prairies
of the Western States towers have had to be built as also
;

has been done in India, where solid towers were used at

first, but were afterwards superseded by hollow ones, which
allowed the instruments to be centred vertically over the
stations. The centres of trigonometrical stations are gen-

erally indicated by a well-defined mark on the upper sur-

face of a block of stone buried at a sufficient distance be-
low the surface. In the Algerian triangulation the stations
were marked by flat-topped cones of masonry having a
vertical axial aperture communicating with the station
mark.

In sunlight, stations may be rendered visible at a great

distance by means of the heliostat, and at night the elec-
tric light is now much employed. In the triangulation
across the Mediterranean already alluded to the signal
lights were produced by steam-engines of six-horse power
working magneto-electric machines. These lights were
placed in the focus of a reflector 20 inches in diameter,

consisting of a concavo-convex lens of glass with the

convex surface silvered. The curvatures of the surfaces
corrected the lens for spherical aberration, and it threw
out a cone of white light, having an amplitude of 24/,
which was directed on the distant station by a telescope.
A refracting lens, eight inches in diameter, was also
used, and threw the light one hundred and forty miles.
There were two Spanish stations fifty miles apart.
Mulhacen, 11,420 feet high, and Tetica, 6,820 feet. The
136 Triangulation.

two Algerian stations, 3,730 and 1,920 feet, were 66 miles

apart, and were each distant from Mulhacen about 170
miles. The labour of transporting the necessary machin-
ery, wood, water, &c, to such a height as Mulhacen was
very great. It was twenty days after everything had been
got ready before the first signal light was made out across
the sea. After that the observations were carried on un-
interruptedly. In France, night observations have been
carried on by means of a petroleum lamp placed in the
focus of a refracting lens of eight inches diameter.

MEASURING THE ANGLES.

Of late years the only instruments used for measuring

the angles of a triangulation have been theodolites of
various sizes the larger natures being really "alt-azimuth"
;

instruments. The more important and extended the sur-

vey the larger and more delicate are the instruments em-
ployed. In the great triangulation of India theodolites
of 18 and 36 inches diameter were used, the average
length of the triangle sides being about 30 miles. For the
Spanish-Algierian triangulation they had theodolites of
16 inches diameter read by four micrometers. In the
United States Coast Survey the large theodolites have
diameters of 24 and 30 inches. For the secondary and
tertiary triangles smaller instruments are used. The
method of taking the angles varies with the nature of the
instrument. The
smaller ones have usually two verniers.
Those of about 8 inches diameter have three, while the
arcs of the larger ones are read by micrometers, of which
some have as many as five. In all cases errors due to
unequal graduation and false centreing are almost entire-

ly eliminated by the practice of reading all the verniers

or micrometers, and taking the same angles from differ-
ent parts of the arc. It is usual to measure all important
angles a large number of times.
 Triangulation. 137

Of the smaller theodolites there are two kinds, the re-

peating and the reiterating. In a repeating theodolite
the lower horizontal plate is free to revolve if necessary,
and thus, when the two plates are clamped at any par-
ticular reading, the telescope can be directed to any point
required. In a reiterating theodolite the lower plate is
fixed to the stand, and when the instrument is set up for
the purpose of measuring a horizontal angle it is quite a
chance what point of the graduation the angle will have
to be measured from. This form is not so convenient as
the other for general purposes, but it has the advantage of
superior stability, and is therefore preferred for running
straight lines. The 6-inch reiterating transit theodolites
used on the Canadian Government surveys have three
verniers. If an angle is read off on each, and the tele-

scope then turned over and the angle remeasured, we

shall have six measures from six equidistant parts of the

arc, the mean of which is taken,. In all cases, after

reading an angle or a round of angles, the telescope
should be again directed on the first object observed, so as
to make sure that the reading is the same and that noth-
ing has slipped.
The method of "repeating" an angle is this. When
the telescope has been directed on the second, or right
hand, object, and clamped, instead of reading the vernier,
the lower plate is set free, and the two together revolved
till the telescope again set on the left hand object.
is

The lower plate then

is clamped, the upper one set free,
and the telescope directed on the right hand object. The
vernier may now be read, when the angle will be half the
reading or the repeating process may be continued for
any number of times, and the whole arc passed over,
divided by the number of repetitions. The object of this
process is to eliminate errors of observation and gradua-
tion. Owing, probably, to slipping of the plates, it does
not usually give such good results as might be expected.
 i3 Tri angulation.

TO REDUCE A MEASURED ANGLE TO THE CENTRE OF

STATION.

It may happen than an inaccessible object such as

the summit of a church spire has to be used as an angle
ofan important triangle. It cannot, of course, be meas-
ured directly, but it may be found indirectly as follows :

Let ABC be the triangle and A the inaccessible point.

Take a contiguous point A' and measure the angles ABC,
BCA, BA'C, AA'B. Calculate or otherwise obtain the
distance AA' on plan.
CallBAC, A; BA'C, A ';

ABA', a and ACA', /?.

;

Now A + a=A'+ /3.

Therefore A^A'+^-a.
Also, AB and AC are
known, and

JAB sin =AA' sin AA'B

JAC /3=AA' sin AA'C
sin Fig. 33.

or, since a and /? are very small angles, if they are taken
in seconds,
(AB x a sin 1" =AA' sin AA'B
(AC x /3sin 1 =AA' sin AA'C
AA' sin AA'B A A' sin AA'C
Therefore, A=A'
AB sin 1' AC -sin i'

CORRECTION FOR PHASE OF SIGNAL.

If thesun shines on a reflecting signal such as a
polished cone, cylinder, or sphere the point observed will,
in general, be on one side of the true signal, and a correc-
tion will have to be made in the measured angle. The
following is the rule in the case of a cylinder.
Let r be the radius of the base of the cylinder, Z the
horizontal angle at the point of observation between the
sun and the signal, and D the distance.
 Triangulation. 139

Then, the correction

r cos
T
D sin 1"

The proof is very simple.

In the case of a hemisphere the value of r will depend
on the sun's altitude. If we call the latter A, r will

become r cos , which must be substituted for r in the

2
above equation.
TO REDUCE AN INCLINED ANGLE TO THE HORIZONTAL
PLANE.

It often happens, as in
the case of angles meas-
ured with the sextant or
repeating- circle, that the
observed angle is inclined
to the horizontal, and a
reduction is necessary to
get the true horizontal
angle. In Fig. 34, let O
be the observer's position,
a and b the objects, andaO b FiS- 34-

the observed angle. If Z is the zenith, and vertical arcs

are drawn through a and b, meeting the horizon in A and
B, then A p is the angle required. aZ b is a spherical
triangle, and by measuring the vertical angles Aa, B6,
we shall have its three sides, since Z A and Z B are each
90. Also, a Z 6= A O B. If we call ab, h ; Za, z; and
Zb, z\ we can obtain a Z b from the equation
. a Z -=
b (sin (s z) sin (s z)
sm
sin z sin z

where s
h+z+z'

The arcs Aa, B6 are generally small, and the differ-

 140 Tri angulation.

ence of z and z therefore also small. The arcs may

therefore be substituted for the sines, and we have for the
correction (in seconds)

A^-O
AOB 7
h = /\go 2 + -'l
Y
2
+
tan
k '

sin r i
Z Z '\
[
2
cot,
k -

sin 1
|

This formula is applicable when z and * are within 3

of 90.

If one of the objects is on the horizon we shall have

AOB-/i= 2 cot h sin 1"

J45 J

If, in addition, the angle h is 90 the correction will be

nil.

THE SPHERICAL EXCESS.

The angles of a triangle measured by the theodolite
are those of a spherical triangle the reason being that at ;

each station the horizontal plate when levelled is tangen-

tial to the earth's surface at that point. must We
therefore expect to find that the three angles of a large
triangle, when added together, amount to more than 180 ;

and this is actually the case. The difference is called the

"spherical excess." From trigonometry we

spherical
know that its amount is directly proportional to the area
of the triangle. In small triangles it is inappreciable.
An equilateral triangle of 13 miles a side would have an
excess of only one second. For one of 102 miles it would
be one minute.

Taking for granted that the spherical excesses of two

triangles are as their areas we can easily find the excess
for a triangle of area s thus A trirectangular triangle
:

It Y2
has a surface of one-eighth that of the sphere, or - ,and
its excess is 90 , or 324000". The excess, in seconds,

will therefore be equal to 2

x s; r and s being,

of course, in the same unit of measure.

 Triangulation .
141

2
Since very small compared with r it may be
s is

obtained with sufficient accuracy for the purpose by

treating the triangle as it were a plane one. We may
thus use either of the formulas
a b sin C
2
a 2 sin B sin C
or
2 sin (B-j-C)
according to the data given.
Approximately, the spherical excess (in seconds) is the
area in square miles divided by 75.5.

The expression for the spherical excess may be put in

the form X
2 r 2 sin 1

For very large triangles we must take the eccentricity

of the earth into account. The expression will then
become
a b sin C (1-f-g
2 cos "it L)
f
2 r 2 sin i'

where e is the eccentricity of the earth, r the equatorial

radius, and L the mean latitude of the three stations.
CORRECTING THE ANGLES OF A TRIANGLE.
In practice the sum of the three measured angles of a
triangle is never what it ought to be, and they have to be
corrected. Supposing that all three angles have been
measured with equal care, the plan adopted is to add to
or subtract from each of the angles one -third of the
excess or deficiency. Thus, if E" was the calculated
spherical excess the three angles ought to amount to
they amounted to i8o +
"
i8o+E". Supposing that
and that n were greater than E. Then we should subtract

from each angle

If some angles have been measured oftener, or with

greater care, than others, the amount of correction to be
142 Triangidation.

applied to each will be inversely as the weights attached

to the results of the measurements.

In the Spain-Algiers quadrilateral triangulation the

spherical excesses of the four triangles were
43".5o; 60". 7; 7o".73; 54".i6
and the errors of the sums of the observed angles were
-fo".i8 o".54 +i".84 +i".i2
CALCULATING THE SIDES OF THE TRIANGLES.
The next step is the calculation of the sides of the
triangles. Treating the latter as spherical this may be
done in three ways.
1. Using the ordinary formulas of spherical trigonome-
try. This is a very laborious method, and others which
are simpler give equally good results.

2. Delambre's method. This consists in taking the

chords of the sides, calculating the angles they make
with each other, and solving the plane triangle thus found.

To reduce an arc a to its chord we have

Chord=2 sin J-
a

or, if the arc be in terms of the radius^

Chord =a -fa as j

The angles made by the chords are obtained by a well-

known problem in spherical trigonometry.
3rd method, by Legendre's Theorem which is, that in ;

any spherical triangle, the sides of which are very small

compared to the radius of the sphere, if each of the
angles be diminished by one-third of the spherical excess,
the sines of these angles will be proportional to the
lengths of the opposite sides; and the triangle may
therefore be calculated as if it were a plane one.

All three methods were used in the French surveys. In

the British Ordnance survey the triangles were generally
c*sr *j-
 Triangulation. 143

calculated by the second method and checked by the

third.

Legendre's theorem gives very nearly accurate results.

In a triangle of which the sides were 220, 180, and 60
miles, the errors in the two long sides, as calculated by
this method from the short side, would be only three-
tenths of a foot.

Thefollowing investigation shows under what circum-

stances small errors in the measurements of the angles of
a triangle have the least effect upon the calculated
lengths of the sides.

Suppose that in a triangle a b c we have the side b as a

measured base, and measure the angles A and C; we have
a sin B=b sin A
If we suppose have been correctly measured we
b to

may treat it and under this supposition

as a constant ;

if we differentiate the above equation we shall get

b cos A . a cos B ,
B
.

da= :
^-dA :
6
-
a
sin B sin B
b a
or, since
sin B sin A
da=a cot A d A a cot BdB
d A B
are here supposed to be positive, and
and d
represent small errors in the measurements of A and B.
If they are assumed to be equal and of the same sign we
shall have for the error of the side a,
da=a d A (cot A cot B)
which becomes zero when A=B.
If d A and d B are supposed equal, but of opposite

signs, we shall have

da= a d A (cot A + cot B)
and since
sin(A + B) sin(A + B)
cot A+cot B A B
sin sin J cos (A B) cos(A + B)
144 Tri angulation.

it follows that
2 sin C
j
da= .
adj A-
a
cos (A B) -I- cos C
and da will be a minimum when A-=B.
In either case we have the result that the best con-
ditioned triangle is the equilateral.
 CHAPTER IV.

DETERMINATION OF THE GEODETIC LATITUDES, LONGI-

TUDES, AND AZIMUTHS OF THE STATIONS OF A
TRlANGULATION, TAKING INTO ACCOUNT
THE ELLIPTICITY OF THE EARTH.

Where the lengths of all the sides of a triangulation

have been computed it becomes necessary, in order to

plot the positions of the stations on the chart, to obtain

their latitudes and longitudes.

The first step to be taken is to determine by means of

astronomical observations the true position of one of the
stations, and also the azimuth of one of the sides leading
from it. Wecan then, knowing the lengths of all the
sides of the trianglesand the angles they make with each
other, deduce the azimuths of all the sides, and calculate
the latitudes and longitudes of the other stations.

Before geodetical operations had been carried to the

perfection they have now attained it was considered suf-
ficient to solve this problem by the ordinary formuloe of

spherical trigonometry, taking as the radius of the earth

the radius at the mean latitude of the chain of triangles.
146 Geodetic Latitudes, &c.

Thus in the triangle PAA'

(fig. 35) where P is the
pole of the earth, and A,
A', two stations, if the
latitude and longitude of
A were known, and also
the length and azimuth of
A A', we should have the
two sides A P, A A', and
the included angle PAA',
and could use Napier's -(Fig. 35.)

analogies to determine the remaining parts of the triangle,

and thus obtain the latitude and longitude of A', and the
azimuth of A at A'. But this method is deficient in

exactness, especially as regards the latitude, and the fol-

lowing has been adopted as giving better results.

Let AN
be the normal at A, and suppose a sphere to
be described with centre N and radius N A meeting the
polar axis at p. Also let p A, p A' be meridians on this
sphere. We
then calculate the geographical position of
A', not by the ordinary formulas of spherical trigonometry

(since the side A A' is very small relatively) but by the

series

I. #=--& c cos A+ c2 cot b sin 2 A

-f-J c
3
cos A sin 2 A (J+cot 2
b)-r...

II. 180 B A + c sin A cot b

+ \ C2 sin A cos -
A (1 + 2 cota b)
+ J c 8 sin A cos 2 A cot b (3+4 cot 2 b)

\ c
s
sin A cot b (1 + 2 cot 2 b)...

III. C= sin b
7 sin A + -.
sin b
: sin A cos A cot b

c c
-f A
9 sin b
-
sin A cos2 A (1 +4^ cot2 b)' ^
a --. ,
sin A cot 2 b.
sin b
 Geodetic Latitudes, &c. 147

Let L
be the latitude of A
" V " A'
" M be the longitude of A =A Pa
" M' " " A' A'P
Let A A' K, and let Z and Z' be the angles it makes
with the meridians pA and pA', respectively. Then, sub-
stituting this notation in the spherical triangle ABC, and
expressing by u the value of in terms of the radius, we K
have
a 90 L b = 90 L
A Z B 180 Z'
c * u C = M' M
which would be the values to introduce into the series
]l.U UT HLp^-H; but in practice it is mor e convenient to count
the azimuths from o to 360, starting at the south and
going round by the west, north, and east. This-makes Z ^k.

the azimuth of A' at A, and Z' the azimuth of A at A'.

*4L_, Therefore in Fig. 35 V=i8o Z, and V'= 3 6o Z', and
the series Irij^jHH-will be changed respectively into
{a) L'= L u cos Z u 2 sin 1" sin 2 Z tan L
...
= M-f-w sin LZ .
7 tan L
(0) M ,

cos
* 2
2 sin 1
.

sin 2 Z-
cos
=-
L
(c) Z'=i8o + Z w sin Z tan L
+i 2 sin 1" sin 2 Z (i-f 2 tan
2
L)
the arc u being supposed to be in seconds.

Pt-^om^tims-4rapperis-ttet the latitude L' is not quite

the true latitude of A'; for the latter is A' N' Ql, or the
angle made by the normal A' N' with N' Q', while the
latitude given by equation (a) is the angle A N Q. The
correction of the latitude {<p) is the angle N' A' N '; for
A 'NQ A 'N 'Q '=A 'RQ '

A.'N 'Q '=N 'A 'R

,
. . N N' sin PN'A'
and sin^= " '

N' A'

Before investigating the exact value of this angle it

should be noted that when the geodesic line K is more
 148 Geodetic Latitudes, &c.

than half a degree its amplitude in latitude on the sphere

say JL becomes a different quantity say AL on

the ellipsoid, and that these two amplitudes of arcs of
the same length being inversely proportional to their radii
of curvature N, R, we have
1 ~* e
A L:rfL::N : R::i : _
1 e
2
sin 2 L
T

whence we have, very nearly

A L ^ L (i+ s cos 2 L), and consequently
<p=d L e cos L
2 2

and therefore the corrected latitude L' is

(OL'=-L cos Z + \u sin

(
2
i"sin 2 Ztan L) (1 + e2 cos 2 L)
and we have in seconds,
K (1 ^ sin 2 L) J K
aa sin 1 N sin 1

The formulas and (c) are not ordinarily used, for when
(b)
the latitude L
known on the spheroid it is used to de-
is

termine M' and Z'. But in this case we must introduce

L' into the values of these two unknown quantities. Now
we have the spherical triangle^) AA', giving
/XT' TVT\
Sm U Sm Z ,,,\
sin (M -M)= cosL (C ) ,

and, since u is very small-,

cos L
Also, in the same triangle
sin (L-f L')
cot J (A + A') tan J (M' M)-cos
(L L')

tan 90
J J

but 90 -
and M' M being

always very small angles, and A + A' being the same as

Z' Z, we have
sin ( L-> ''
(c) Z-i8o'+Z(M'-M) cos .*
T
(L L)
 Geodetic Latitudes, &c. 149

The imaginary sphere used in the above investigation

will, of course, coincide with the spheroid for the parallel

of latitude through the point A. Any plane passing
through the normal will cut the surface of the sphere in
the arc of a great circle, and the spheroid in a line, which,
for about three degrees, will be practically a geodesic
line.

Thefollowing is another way of treating the sub-

ject. Instead of taking the normal at one of the points
A A' as the radius of the imaginary sphere let us take the
normal at the point B, mid-way between them, as in Fig.
36, and for the sake
of simplicity let these
points be on the same
meridian. Let A N,
A' N' be the normals
at A A', produce them
to Z and Z' respec-
tively, and draw A e,
A' e parallel to the

major axis O E.
The astronomical lati"
tudes of the two points
are Z A e, 71 A' e If .

now we draw B C the Fig j6

normal at B, C will fall between N and N\ The curve
given in the figure is the elliptical meridian. The circular
curve drawn with radius C B is not shown but it would ;

pass a little outside of A and A'. For practical purposes

we may suppose it to pass through those points. Join
C A, C A', and produce them to z and z respectively.
z A e, z A' e' will be the latitudes of A and A' on the

imaginary sphere, one being less and the other greater

than the latitudes on the spheroid. The differences
Z A 2, 21 A' z may be considered the same. Let each be
150 Geodetic Latitudes, &c.

designated- . Let L and L' be the astronomical lati-

tudes of A and A', /, and /' their latitudes on the sphere,

and X the latitude of B. Then #=L L' (/-/')
, L + L' / + /' ,
or ,
and
2 2
/ /' radius of curvature at B 1 e2

L L normal at B 1 e 2 s\i\ 2 X

A1 I I' L L' d 1
Also
L L' L L' 1+62 COS2 X

therefore
L L'
L L'-
+ e 2 COS 2
1 /

and o*=(L -
M
i+ 2 cos 2 /r
L') -(
1

' 2COs2;
= (L-L') 1+e 2 cos 2
/

(L L')
2 cos 2 Xy nearly.
The angle <5 is therefore nearly the same as the correc-
tion <p already investigated.
In what next follows K is the distance A A' in yards
of any two stations A, A', u the same distance in seconds
of arc, R the radius of curvature of the meridian, N the
normal (both in yards), e the eccentricity (=0.0817), and
a the equatorial radius.

Equation {a) gives us the values of u and L', (b') gives

us M', and (c) gives Z'. If we neglect the denominator
of the fraction in (c) we have
Z'=i8o" + Z (M' M) sin J- (L+L')
c
or Z'=i8o + Z ^-5HLf s j n 1 (L+L')
cos L
The last term of this equation, which is the difference
of the azimuths at the two stations, is the convergence of
their meridians.
 Geodetic Latitudes, &c. 151

If the triangulation is limited in extent it may be more

convenient to express L', M
Z' in terms of rectangu-
',
and
lar co-ordinates referred to axes having their origin at the
station A, the axis of y being the meridian at A, and the
axis of x the geodesic line through A and perpendicular
to the meridian. The equations are

*
L'=L =r-4!
R sin i"
Jsin
*
i4
(N sin i"j
Kan /L
R
4
sin
)
V V)
M'=M ^-4
N sin i"
x ^r-
cosL*

Z'=i3o + Z-^4
N sin 1
^tan L'*

The sphere described with radius equal to the normal

for the mean latitude of two stations, A and B, may be
used in the next three problems.

I. GIVEN THE LATITUDES AND LONGITUDES OF TWO

POINTS TO FIND THE LENGTH AND DIRECTION OF
THE LINE JOINING THEM.
Here we have given L, L', M, and M ',
and from
L and L' we obtain d.

We have then to find / and /' from the equations

/_ L _ jL
2
and /' L'-f
2

Let x" be the number of seconds in the arc passing

through the point of which L is the latitude and perpen-
'

dicular to the meridian through the other point.

Lety be the number of seconds in the portion of this

meridian between L and the foot of this perpendicular.

Let x, y, be the same quantities in linear units, N=the

normal at the middle latitude, and Z==the azimuth of the
point y/ from Jfa.
/\' ^4 _

"These three formulae, as well as the formulas and examples given in

the next few pages, have been taken from Jeffers' Nautical Surveying.
Fractions of seconds have been omitted.
152 Geodetic Latitudes, &c.

Then we shall have

** (M' M)cos/'
y l /'
i sin 1" x** tan /

** x N sin i"

jy==y" N sin I"

tan Z=
y"
x" y'
/<

sin Z cos Z
K=w" N sin i"

The signs of (L L') and of (/ /') must be carefully at-

tended to.

EXAMPLE.
Given
L=49
4' 25"
22 33
L'=49
M' M, or difference of longitude-=38' 47"=2327"
to find Z and K
Here L + L'-= 9 8 26' 58"
A = J(L+Ii) 49 13 29
L' L= o 18 8
KL/-LJ-0 9 4544*

To find the value of

2
log 027.81085
log J (L'L) 2.73549
2 log cos J (L -f L)=9.62994

(J

Jog 0.17628

-1 .5

J L =49 4 26"'.5 ( being negative)

f'=L' +
/

2
=4922 3i".5
 Geodetic Latitudes, &c. 153

To find x"

Log (M'M) =3.3668785

log cos 11=9. 8136470

log ^=3.1805255
*"=I5I5"
To find the value of the 2nd term of y"
'

log i sin 1 '=4.38454

2 log x' '=6.36105
log tan /=o. 06197

log 2nd term=o. 80756

2nd term=o o' 6"

I'lo 18 5

y"=o 18 1 1-^1091"
To find the azimuth Z
Log *"=3. 1805255
log/^3.0378887
x
log 7T =0.1426368
Z=i2545' 21"

To find log N sin 1"

Log N (in yards) = 6.8443224

log sin i/'=4.6855749

Log N sin 1"= 1.5298973

To find log u"

Log/^3.0378887
log cos Z=9. 7666596

log ^"=3.2712291
To find K
log ^"=3.2712291
log N sin i" = i. 5298973
4.8011264
#=63226 yards,
154 Geodetic Latitudes, &c.

To find the co-ordinates.

Value of x. Value of y.

Log *"=3. 1805255 Log ^"=3.0378887

log N sin 1"= 1.5298973 log N sin i"=i. 5298973

log ^=47104228 log ^=4.5677860

#=51336 yards ^ = 36965 yards

TO COMPUTE THE DISTANCE BETWEEN TWO POINTS,

KNOWING THEIR LATITUDES AND THE AZIMUTH OF
ONE FROM THE OTHER.

Let L and L' be the latitudes, Z the azimuth, and let

(L+L') _^
2

Then we shall have, as before

d __ 2
e (L L') cos X
2
a
N 2
(i~e sin 2 X)i

/-l ^ r=L'+-
2 2
. tan /
Assume -=tan <p
cosZ
.
/ t, s sin /' .

then, sin (<p

r u )
sin
r sin
/
cc

which gives u; and K w" N sin i

#/

The algebraic sign of cos Z will determine the sign of <p,

and, consequently, whether u" is to be added to or sub-

tracted from <p.

Example
L= 49 4' 25" N
L' 49 22 33 N
Z = 125 45 21
Here, as in the last example, we find ", and hence

/=49 4' 27"

/'=49 22 32
 Geodetic Latitudes, &c. 155

To find the value of f. To find <p

u.
=
log tan l=o. 0619727 log sin ^ 9-95 3895
=
log cos Z g. 7666566 log sin l'~g. 8802377
co-log sin Z=o.i2i7320
log tan ^ = 0.2953161
?=-(637_'55") log sin (<p-u) = g.g523*>92
<pu=(63 3 g 3")
u=o* 31 8"= 1868'
From the equation K= N sin 1" we find K to be
63228 yards.
TO COMPUTE THE DISTANCE BETWEEN TWO POINTS,
KNOWING THE LATITUDE OF ONE, THE AZIMUTH
FROM IT TO THE OTHER, AND THE DIFFERENCE
OF THEIR LONGITUDES.

Using the same nomenclature as before, let L be the

given latitude and m the difference of the longitudes:
Take L" = L' +
Assume tan ^ = sin L tan Z
tan L" = tan L oin (f mf j^-^
d=e 2 (L L")cos
2
i(L + L"), very nearly
V=V'-d
2

2
m cos /
K=u" N sin I
1

sin Z

The algebraic sign of tan Z will determine the sign of <p s

and consequently whether it is to be increased or dimin-
ished by m.

Example-
Let L= 49 4'25',/N
Z-x'25 45 21
m=3R' 47"=2327"
156 Geodetic Latitudes, &c.

To find <p. To find L'

log sin 1^=9.8782652 log tan L=o.

L=o.c0619663
log tan Z=o. 1426368 g sin (<p
log m)=g. 8643024
=
co-log sin ^ 0.1403154
log tan <p
= o. 0209020
<p= (46 22 42") log tan L"=--o. 0665841
m- 3 8 47 L"=49*22'3o"

^3
 Geodetic Latitudes, &c. 157

On the North American boundary survey in 1845 the

following method was employed to find the mutual azi-
muths of two distant points the latitudes and longitudes
of which were known.

Let A and B be the two points, of which B is the

northern, and P the pole. Then, treating the earth as a
sphere, we have in the spherical triangle P A B the two
sides PA, P B, and the angle A P B given, and have to
find the angles A, B. This is done by the usual formulae,

cos
AP-BP
tanHA+B) * cot-
A p^ B p
cos
2

. AP-BP
sin
X COt
tan I (B-A)-
sin
XpTBP T
2
.. . . A+B , B
-
A
which give and
2 2
B B-A
Then, A: 2
2

B= A+B B-A
2 2

To correct these azimuths for the earth's spheroidal

form take 90 A and B go, and calculate the angles
a, /?, from the formulae
sin AP .
Q BP
sin
sin . sin p=-
V 75 ^75
Then, if A' and B' are the true spheroidal azimuths,
tan (90 A')=cos a tan (90 A)

tan (B go)=cos P tan^B-90 )

This method is very useful when a long line has to be

cut from one point to another through forests.
158 Geodetic Latitudes, &c.

To find the accurate length of the arc on the surface of

the earth between two very distant points of known lati-
tude and longitude is a very difficult and not very useful
problem. It is, however, often advisable to calculate the
distances between stations that are within the limits of
triangulation, as a check upon the geodesical operations ;

and in the case of an extended line of coast, or in a wild

and difficult country where triangulation is impossible,

thisproblem is most useful for the purpose of laying down

upon paper a number of fixed points from which to carry
on a survey.
In the triangle PAB mentioned in the last article we
have, as before, the sides PA, PB, and the angle P, as
data. By solving the triangle we obtain the length of the
arc AB. If the azimuths can be observed at the two
stations the accuracy of the result will be greatly increas-
ed, and we can obtain the difference of longitude of the
two stations as follows be proved that the sphe-
: It may
rical excess in a spheroidal triangle isequal to that in a
spherical triangle whose vertices have the same astrono-
mical latitude and the same differences of longitude :

from whence results the rule

PA-PB
-r, COS . .
-r,
P = 2 ^A+B
tan ^ x cot
2 PAA -f -rt^
,

PB 2
cos :

2
cos diff. lat. A+B
x cot-
sin J sum of lat. 2

which gives P, or the difference of longitude.

As a rule, a small error in the latitudes is of no import-
ance unless the latitudes are small but the azimuths :

must be observed with the greatest accuracy. The angle

P being known we can get the length of the arc AB, and
must then convert it into distance on the earth's surface,
using the radius of curvature of the arc for the mean lati-

tude.
 Deville' s Methods. 159

It be observed that when dealing with the sphere

may
we have the definite equations of spherical trigonometry
to work with while, when a spheroidal surface is in
;

question, we have only approximate formulae. In most

of the equations employed in higher geodesy the right
hand side consists of the first few terms of a converging
series, the remainder being so small that they may be
omitted without causing any appreciable error. Thus, in

calculating the length of a geodesic line between two dis-

tant points the smaller terms of the series might amount
to only a few inches in 100 miles.

Captain Deville in his "Examples of Astronomic and

Geodetic Calculations" gives some very simple methods
of solving certain problems in Geodesy by means of
tables of logarithms of the convergence of meridians for
O^o o$r chain departure, and tables of the value of a chain
in seconds of latitude and in seconds of longitude at
different latitudes. By departure is meant, of course,
the distance one point is east or west of the other. If a

great circle (not being a meridian or the equator) is drawn

on the earth's surface it will cut each meridian it crosses
at a different angle according to the laitude of the point
of section: in other words its azimuth is continually

changing and if we take two points, A and B, on this

;

great circle, and P is the pole, the convergence between

A and B will be, practically, 180 (PAB + PBA.) If the
two points are in the same latitude, and one chain dis-
tant from each other, each of the angles A and B will be
lessthan 90 by half the convergence. If the distance be
constant the convergence will increase as we recede from
the equator (where it is nothing) towards the poles. In
problems involving two stations of different latitude the
convergence used it that for the mean of the two latitudes.
The examples given by Captain Deville are worked out
by logarithms. In the following selection of problems
the principle only of the method is indicated.
160 DevilWs Methods.

Prob. i. To find the convergence of meridians between

two points of given latitude.

Here we have only to find by a traverse table the de-

parture in chains and multiply it by the convergence for
one chain for the mean latitude.

Prob.
2. To refer to the meridian of a point B an
azimuth reckoned from the meridian of another point A.

Calculate the convergence between the two points, and

add or subtract it from the given azimuth according as B

is east or west of A.

Prob. Given the latitude and longitude of a station

3.
A, and the azimuth and distance of another station B, to
find the latitude and longitude of the latter.

The
distance and azimuth being given we can find the
departure and distance in latitude of B approximately by
the traverse table, and have the approximate mean lati-
tude. We
next find the mean azimuth by multiplying
the departure by the convergence for one chain at the
mean and applying the convergence thus ob-
latitude,
ained to the azimuth of B at A, which gives the azimuth
of A at B, and hence the mean azimuth.

get the correct latitude of B we multiply the dis-

To
tance by the cosine of the mean azimuth and by the value
of one chain in seconds of latitude. This gives the differ-

erence of latitude of the two stations in seconds.

Similarly, the difference of longitude of the stations is

found by multiplying the distance by the sine of the mean
azimuth and by the value of one chain in seconds of
longitude.
Prob. 4. To correct a traverse by the sun's azimuth.

On a traverse survey of any extent the direction of the

lines must be corrected from time to time by astronomi-
cal observations, usually either of the sun or the pole star.
 Devilles Methods. 161

If the traverse is commenced at a station A with a known

orientation, and at another station B an observation is
taken, the azimuth or bearing of the line thus obtained
should differ from the azimuth as carried on through the
traverse from A by the convergence of meridians between
the stations. Should there prove to be any error it should
be equally distributed among the courses by dividing it by
the number of stations. Multiplying the result by the
number of any course gives the correction for that course.
As an example: A traverse of seven courses was made in
a westerly direction. At station 8, or the end of the 7th
course, the sun was observed, and its azimuth found to be
7
^fe- 11' 50", the horizontal plate reading being 267 59' 10".
Required the error of grientation.
On calculating the convergence between stations 1 and
8 it was found to be 49 5*. If the traverse had been cor-
'

azimuth plus the convergence would

rectly run the sun's
have been the same as the plate 207-
reading; but the latter was 1 45"
too little. One seventh of this,
or 15", the correction for each
is

course, and we have to add 15"

to the plate reading of the first
course, 30" to that of the second, 1 45
and so on.
Prob. 5. When running a line to correct its direction

by the sun's azimuth.

Unless the line is a north and south one its azimuth
will be continually changing from point to point. Its
direction can be checked at any time by finding its azi-
muth astronomically to ascertain if this is what it ought
to be after allowiug for the convergence. The first step is
to find the approximate difference of latitude from the
distance chained and the azimuth at which the line
started. This will give the latitude of the station and
1 62 Deville's Methods.

the mean latitude approximately. The latter being

known, the azimuth and distance give the convergence,
which being applied to the initial azimuth the true azi-
muth is obtained.

Prob.
6. To lay out a given figure on the ground, cor-
recting the courses by astronomical observations.
Take as an instance a square ABCD, the side AB
being commenced at A with a given azimuth. The course
is to be corrected by observations at the other three corners.
The convergence between A and B being found in the
usual manner and applied to the original azimuth (in
addition to the angle at the corner) gives us the azimuth
of BC. Similarly, the convergence between A and C will
give us the azimuth of CD ;
and so on.
Prob. 7. To lay out a parallel of latitude by chords of
a given length.

The angle of deflection between two chords is the con-

vergence of meridians for the length of a chord, and the
azimuths will be 90 minus half the convergence and 270*
plus half the convergence. The convergence is found in
the usual way.

Prob. 8. To lay out a parallel of latitude by offsets.

A parallel may be laid out by running a line perpendicu-

lar to a meridian and measuring offsets towards the near-
est pole. The length of an offset is its distance from the
meridian multiplied by the sine of half the convergence
for that distance; or (since the distance is in this case the
same as the departure) the square of the distance multi-

plied by the sine of half the convergence for one chain.

As this angle is small the logarithm of its sine is ob-
tained by adding the logarithm of the sine of half a second
to the logarithm of the convergence for one chain de-

parture.
When the offsets arc equidistant any one of them may
 DevilWs Methods. 163

be obtained by multiplying the first one by the square of

the nnmber of the offset.

It is almost superfluous to point out that in practice all

these problems are worked out by means of logarithms.

TO FIND THE AREA OF A PORTION OF THE SURFACE OF

A SPHERE BOUNDED BY TWO PARALLELS OF LATI-
TUDES AND TWO MERIDIANS (SPHERICAL SOLUTION.)
Let AB and CD be the meridians and AC, BD the
parallels. Let <p
be the latitude of A, <p' of B and n the
difference of longitude of the meridians.

Now the area of

the whole portion
of the surface com-

prised between two

parallels is equal to
the area of the por-
tion of the circum-

scribing cylinder Flg 37

(the axis of which is the polar axis) contained
between
the planes of the parallels produced to meet it. {Vide
second figure showing a section, in which a is the point A
and b the point B.)

Let r be the radius of the sphere and h the perpendicu-

lar distance between the planes.

Then the area of the spherical zone will be

2 tz rxh
=2 re r xr (sin <p'
sin tp)

=2 7T r 2 (sin <p'
sin <p)

;. the area of the portion between

the two meridians
will be

n 7T y
(sin f sin <p)
180
164 Offsets to a Parallel.

TO FIND THE OFFSETS TO A PARALLEL OF LATITUDE,

Let PA, PBC, be meridians, AB

a portion of the parallel, AC a por-
tion of a great circle touching the

parallel at A.

It is required at a given latitude to

find the offset BC for a given dis-
tance AC.

Fig. 38.
Let x be the circular measure of AC
"
y do. do. BC
"
/ do. do. PA
AC andBC are very small.
In triangle PCA we have cos PC=cos / cos x
x2
=cos /
(1 ) nearly,
x2
Therefore cos / cos PC=cos /
2

or 2 sin

or 2 sin
.

/
v
22
/-f-PC

=x
.

sin

2
-BC

cos
=cos ,
/
x*
2

/, nearly.

x 2
therefore, y=i cot /.

(or, if x andjy are measured lengths, and R is the radius

x2
of the earth, Jv= ^ cot/)
2 R
Next, join AB by a great circle arc. The angle B AC will
be half the convergence, and AB=AC,
approximately.
Draw PD bisecting P, and therefore at right angles to AB.
In the triangle APD we have D 90*
COnvergence
andFAD- 9o- .

2
 Offsetts to a Parallel. 165

Therefore, cos PAD=tan AD cot /,

AD
'

or sin cot /, approximately

=J x cot / ,,

x 2 cot
.
convergence
s
Therefore, y=f / x sin

This is equally true if x and jy

are measured lengths.
 CHAPTER V.

METHODS OF DELINEATING A SPHERICAL SURFACE ON A

PLANE.

Since the surface of the globe is spherical, and as the

surface of a sphere cannot be rolled out flat, like that of
a cone, it is evident that maps of any large tract of coun-
try drawn on a flat sheet of paper cannot be made to ex-
actly represent the relative position of the various points.
It is necessary, therefore, to resort to some device in order
r
that the points on the map may have as nearly as possible
the same relative position to each other as the corres-
ponding points on the earth's surface.

One method to represent the points and lines of the

is

sphere according to the rules of perspective, or as they

would appear to the eye at some particular position with
reference to the sphere and the plane of projection.
Such a method iscalled a projection. The principal pro-
jections of the sphere are the "orthographic," "stereo-
graphic," "central or gnomonic" and "globular."

A second method is to lay down the points on the map

according to some assumed mathematical law, the con-

dition to be fulfilled being that the parts of the spherical
surface to be represented, and their representations on the
map, shall be similar in their small elements. To this
 Projections. 167

class belongs Mercator's Projection, in which the meridians

are represented by equi-distant parallel straight lines, and
the parallels of latitude by parallel straight lines at right
angles to the meridians, but of which the distances from
each other increase in going north or south from the
equator in such a proportion as always to give the true
bearings of places from one another.
The third method is to suppose a portion of the earth's
surface to be a portion of the surface of a cone whose
axis coincides with that of the earth, and whose vertex is
somewhere beyond the pole, while its surface cuts or
touches the sphere at certain points. The conical sur-
face is then supposed to be developed as a plane, which
it of course admits of being. The only conical develop-
ment that will be discussed in these pages is the one known
as the "ordinary polyconic."
The Orthographic Projection is simply the one employed
in plans and elevations. When used for the delineation
of a spherical surface the eye is supposed to be at an in-
finite distance, so that the rays of light are parallel, the

plane of projection being perpendicular to their direction.

In the case of a sphere the plane of projection is usually
either that of the equator or of a meridian. When a hemi-
sphere projected on its base in this manner the relative
is

positions of points near the centre are given with toler-

able accuracy, but those near the circumference are com-
pletely distorted. The laws of this projection are easily
deduced. Amongst others it is evident that in the case

of a hemisphere projected on
its base all circles passing

through the pole of the hemisphere are projected as

straight lines intersecting at the centre. Circles having
their planes parallel to that of the base are projected as

equal circles. All other circles are projected as ellipses

of which the greater axis is equal to the diameter of the
circle, and the lesser axis to the diameter multiplied by
the cosine of the obliquity.
 1 68
Projections.

Stereographic Projection. In this projection the eye is

supposed to be situated at the surface of the sphere, and
the plane of the projection is that of the great circle
which is every where go degrees from the position of the
eye. It derives its name from the fact that it results from
the intersection of two solids, the cone and the sphere.
Its principal properties are the following: i. The pro-

jection of any circle on the sphere which does not pass

through the eye is a circle; and circles whose planes
pass through the eye are straight lines. 2. The angle

made on the surface of the sphere by two circles which

cut each other and the angle made by their projections
are equal. 3. If C is the pole of the point of sight and c
itsprojection; then any point A is projected into a point a
such that c a is equal to

-2.

where r is the radius of the sphere. From the second

property it follows that any very small portion of the
spherical surface and its projection are similar figures a ;

property of great importance in the construction of maps,

and one which is also shared by Mercator's projection.
The astronomical triangle PZS can evidently be easily
drawn on the stereographic projection. Z will be the
pole of the point of sight. The lengths of ZP and ZS
are straight lines found by the rule given above, and the
angle Z points P and S are known.
being known the
The angles P and S being also known we can draw the
circular curve PS by a simple construction.

The orthographic and stereographic projections were

both employed by the ancient Greek astronomers for the
purpose of representing the celestial sphere, with its
circles, on a plane.
Gnomonic or Central Projection. In this case the eye is

at the centre of the sphere, and the plane of projection is

 Projections. 169

a-plane touching the sphere at any assumed point. The

projection of any point is the extremity of the tangent of
the arc intercepted between that point and the point of
contact. As the tangent increases very rapidly when the
arc is more than 45 and becomes infinite at go", it is evi-
,

dent that this projection cannot be adopted for a whole

hemisphere.
Globular Projection. This is a device to avoid the dis-
tortion which occurs in the above projections as we
approach the circumference of the hemisphere. In the
accompanying figure let AC B
be the hemisphere to be repre-
sented on the plane A B, E
the position of the eye, O the
centre of the sphere, and EDOC
perpendicular to the plane A B.
M and F are points on the
sphere, and their projections are
N and G. Now the representa-
tionwould be perfect if A N :

N G G O were as A M M F
: : :

F C. This cannot be obtained FiS- 39.

exactly, but it will be approximately so if the point E

is so shown that G is the middle point of A O and F the
middle point of A C. In this case, by joining F O and
drawing F L perpendicular to O C, it may easily be shown
that E D is equaj to O L, which is O F x cos 45 or ,

r x 0.71 nearly beeaasejffcfls half the radius and F L

half the inscribed square therefore.
FL G : 0::0 C:OL
but F L :G Or:L E O E :

;.LE:OE::OC:OL
consequently, LO O E C L O: :: :
L, or O L 2 =0 E.C L
but O L 2 =F L 2
=D L. L C, /. O E.L=DC L. C L
or O E=D L
that is, E D O L
 170 Projections.

The above projections are seldom used for delineating

the features of a single country or a small portion of the
earth's surface. For this purpose it is more convenient
to employ one of the methods of development.
Mercator's Projection is the method employed in the
construction of nautical charts. The meridians
are repre-
sented by equi-distant parallel straight lines, and the
parallels of latitude by straight lines perpendicular to the
meridians. As we recede from the equator towards the
poles the distances between the parallels of latitude on the
map are made to increase at the same rate that the scale
of the distance between points east and west of each
other increases on the map, owing to the meridians being
drawn parallel instead of converging. If we take / as the

length of a degree of longitude at the equator (which

would be the same as a degree of latitude supposing the
earth a sphere), and /'that of a degree of longitude at
latitude k, then /'/cos k, or /'--/::i : sec A. Now
V :the proportion in which the length of a given dis-
I is

tance in longitude has been increased on the map by

making the meridians parallel, and is therefore the pro-
portion in which the distance between the parallels of
latitude must be increased. It is evident that the poles

can never be shown on this projection, as they would be

at an infinite distance from the equator.

If a ship steers a fixed course by the compass this

course is always a straight line on a Mercator's chart.

Great circles on the globe are projected as curves, except

in the case of meridians and the equator.

In this projection, though the scale increases as we

approach the poles, the map of a limited tract of country
gives places in their correct relative positions.

The Ordinary Polyconic Projection. In conical develop-

ments of the sphere a polygon supposed to be inscribed
is

in a meridian. By revolution about the polar axis the

 Projections. 171

polygon will describe a series of frustums of cones. \(

the arc of the curve equals its chord the two surfaces will
be equal. In this manner the spherical surface may be
looked upon as formed by the intersection of an infinite
number of cones tangential to the surface along succes-
sive parallels of latitude. These conical surfaces may be
developed on a plane, and the properties of the resulting
chart will depend on the law of the development.

The Ordinary Polyconic is a projection much used in

the United States Coast Survey. It is peculiarly appli-

cable to the case where the chart embraces considerable

difference in latitude with only a moderate amplitude of

longitude, as it is independent of change of latitude.

Before describing it it must be noted that whatever
projection is used the spheroidal figure of the earth must
be taken into account, its surface being that which would
be formed by the revolution of a nearly circular ellipse
round the polar axis as a minor axis.
In the Ordinary Polyconic each parallel of latitude is
represented on a plane by the development of a cone
having the parallel for its base, and its vertex at the
point where a tangent to a meridian at the parallel in-
tersects the earth's axis, the degrees on the parallel pre-

serving their true length. A straight line running north

and south represents the middle meridian on the chart,
and is made equal to its rectified arc according to scale.
The conical elements are developed equally on each
side of this meridian, and are disposed in arcs of circles
described (in the case of the sphere) with radii equal to
the radius of the sphere multiplied by the cotangent of
the latitude. The centres of these arcs lie in the middle
meridian produced, each arc cutting it at its proper
latitude.

These elements evidently touch each other only at the

middle meridian, diverging as they leave it. The curva-
172 Projections.

ture of the parallels decreases as the distance from the

poles increases, till at the equator the parallel becomes a
straight line.
To trace the meridians we set off on the different

parallels (according to the usual law for the length of an

arc of longitude) the true points where each meridian
cuts them, and draw curves connecting those points.

To allow for the ellipticity of the earth we must use for

the radius of the developed parallel N cot /, where
N= e2 sin 2
(i /)

a being the equatorial radius, e the eccentricity, N the

normal terminating in the minor axis, and / the angle it
makes with the major axis.
It is evident that
the slant height of the
cone say r is N cot U
and that the radius
of the parallel on the

spheroid is N cos /.

The length of an arc

of n of a parallel will

be n
180
-
c z--z N cos /.
H
In practice, instead Fig- 4.

of describing the arcs of the parallels with radii, it is more

convenient to construct them from their equations as
circles. The intersections of the meridians and parallels
can also be found in this way. Express x and y, the rec-
tangular co-ordinates of a point, as functions of the
radius of the developed parallel (N cot /) and the angle
(0) that this radius makes with the middle meridian.
 Projections. *73

Take the origin at L

(Fig. 41) the point of in-
tersection of a parallel
with the middle meridian;
the middle meridian as
the axis of y; and the per-
pendicular through L as
the axis of x. Then we
shall have for any point P
whose latitude is / and
longitude from the meri- Fig. 41.
dian n
x=Y P sin 0=N cot sin /
(1)

y=Y P versin 0=N cot versin /

(2)

6 being, of course, some function of n.

To find the relation between and n, since the parallels

are developed with their true lengths the distance LP
equals the length of the portion LP of the parallel on the
spheroid. Therefore the angles at the centres of the two
arcs will be inversely proportional to the radii, and

N cot l_ n' or 0=n sin /

(3)
N cos l~ ~J
These three equations are sufficient to project any
point of the spheroid when we know its latitude and its

longitude from the middle meridian. If we take n con-

stant we can project the successive points of any meri-
dian.

If S is the distance on the elliptical middle meridian

from the origin to the point where the parallel through
the point to be projected cuts the middle meridian, equa-
tion (2) will become, jy=N cot / versin #S.
174 Projections.

From the above equations

tables may be formed for the
construction of charts.

Fig. 42 shows the geometri-

cal relation between the angles
ti and //.

This projection, when the

amplitude longitude does not
in

exceed three degrees from the

middle meridian, has the fol- Fis- 4 2

lowing properties.
It distorts very little, and has great uniformity of scale.
It is well adapted to all parts of the earth, but best to
the polar regions.

The meridians make practically the same angles with

each other and with the parallels as on the sphere. Angles
are projected with little change.

The great circle or geodesic line is projected as a

straight line practically equal to itself.
 CHAPTER VL

TRIGONOMETRICAL LEVELLING.

TO FIND THE HEIGHT OF A POINT B ABOVE A STATION A.

In the accompanying figure

O is the centre of the earth,
AC is tangential to the earth's
surface at A, B' is the apparent

position of B, owing to refrac-

tion. CC is the correction for

K2
curvature, or ^ where K is
2 R ,

the horizontal distance of B

from A, and R is the radius of FiS- 43-

the earth; both in feet. BB' is about 0.16 CC

ACB may be taken as a right angle, and AC the arc ,

AC, and the straight line AC', are all^qualT^vVe shall

have then, if-trre-distancc K is not very gre a t, _,

BC=K tan B'AC + CC-BB'

=K tan B 'AC + 0.00000002 K 2

where B'AC is the observed angle of elevation of B. This

formula supposes that AC B is practically 90. If the dis-
 176 Trigonometrical L evelling.

tance is so great that this is not the case we shall have

in the triangle ACB
BC=K smBAC
sinB
To find the angle B, we have in the triangle AOB,
B=i8o-(0+BAO)
180 (0 + go + BAC)
90 (O + BAC)
Hence, sin B cos (O+BAC)
BAC sin
BC=Kx + BAC)cos (O
K3
And BC'~= BC + CC BC + ^
2K
(where R is the radius of the earth in feet.)

The angle O, in minutes, is 0.0001646 K, and

is 0.000000023936 K2
2R
RECIPROCAL OBSERVATIONS FOR CANCELLING REFRAC-
TION.
If we measure
the reciprocal angles of elevation and de-
pression stations
of two
in other words, if at each

we observe the zenitji distance of the other we shall get

rid of the effects of refrac-
tion. Let a be the angle
of elevation of B at A and
fi the angle of depression
of A at B.

Then
sin + /?)
BC'=Kx (

cos i (+/?+0)
If the zenith distances
are observed call them d and
d', and we shallhave (since
90 a and d'=go + ft)
 Trigonometrical Levelling. 177

sin i (d'S)
BC'=Kx
If O is very small compared with the other angles we
may neglect it, when we shall have
BC'=K tan J (a+p)
K tan J- (P8)*
REDUCTION TO THE SUMMITS OF THE SIGNALS.

Suppose there are two stations, a and 6, which cannot

be seen from each other, so that
signals have to be erected at
each. Let A and B be the sum-
mits of the signals, a and /? the
true angles of elevation [and de-

pression of a and b respectively.

At a the angle B a C is observed
and at b the angle A b D. Call
BaC,d; AbD,</>; A a, h and ;

B b, Then, to find the re-

h'.

duced angles a and ft we shall

have

cos^ #
h
6 _
Ksin 1

h' cos ip
p-?+ K sin 1" Fig- 45-

the differences being: in seconds. ,

If zenith distances A and A '

are taken we shall have

*Clarke gives the formula

h'h=K tan i (d'd)

ji+^tA-1
where h and h' are the heights of the stations, and r the
radius of the earth.
178 Trigonometrical Levelling.

for S and d
= h sin A
A+ K sin 1"

h' sin A '

*=A'- K sin 1"

Reciprocal observations ought to be simultaneous in

order that the effects of refraction may be as nearly as
possible the same for both.

In problems of this kind we ought, strictly speaking,

instead of using the mean radius of the earth, to take the
normal for the mean latitude of the stations.

The following geodetical formulae are used for more

exact determinations. In addition to the letters used in
the foregoing problems we have a the known altitude of
the lower station N the normal for the mean latitude ;
;

M the modulus of common logarithms; and r the co-

efficient of refraction.

1. TO FIND THE DIFFERENCE OF LEVEL BY RECIPROCAL

ZENITH DISTANCES.

Log. diff. oflevel=log {k tan i &-rd)\

2. TO FIND THE DIFFERENCE OF LEVEL BY MEANS OF A

SINGLE ZENITH DISTANCE.

Log. diff. level = log

M M
+,

XT
*
,

1 2 r
'

12 N2
I

2 N sin I 1

The third term is positive if a is less than 90 .

 Trigonometrical Levelling. 179

3. TO ASCERTAIN THE HEIGHT OF A STATION BY MEANS OF

THE ZENITH DISTANCE OF THE SEA HORIZON.
In this case, when possible, different points of the hori-
zon should be observed on different days and the mean
of the whole taken, the state of the tide being also noted.
The formula is
"
Sm X
2
Log. altitude=log 2 j I r
-I'+log. (d 90 )
\ )

. M f sin i" I
s . -

**~
The angle d 90" is in seconds (^-=-^0 )

The last term may generally be neglected.

The following is an example of finding the difference of
level by a single zenith distance.
The altitude of the lower station (a) was 1000 yards,
and h or the height of the instrument 5.

The
horizontal distance between the stations (K) was
57836 yards. The zenith distance of the upper station
(A) 88 24' 40".
First, to find the value of the angle d.
Log h=o.6g8gy
Log sin A =9.99984
Co-log K=5. 23780
Co-log sin i"=5.3i443

T A =I
A sin ,

g '
25I 4 g X7*'
Ksm7*
Therefore (5=88 24 57".8

Next, to find the value of the angle r^ : ,-K

?

2 N sm 1

Log 2
1
=-=
N
2 r
sin
;

1
7l
= 8.13252 D

Log K =4.76220

2.89472=log 784".7=ro 13' 4^.7

2r
* \t
- K=88 n' JJ
53".i
2 N sin 1
i8o Trigonometrical L evetling.

Thirdly, value of the difference of level.

Log K =4.7621984
Log tan 88" n' 53".! =1.5022427

Log 1st term=3. 2599557

2nd term= 691
3rd term= +627
4th term= 9

Log. diff. level=3*26oo884=log 1820.07 yards.

Second Term. Third Term. Fourth Term.

= ^ =2-5383 M
Log *- Log Log
N
2-8393
2N TTnT =34387
Log a =3 log 1 st term =3-2599 log K2 =95244

Log 2nd term = 5-8393 log 3rd term =57982 log 4th term = 2963i

2nd term =0.0000691 3rd term =00000627 4th term =0-0000009

REFRACTION, <fcc.

TO FIND THE CO-EFFICIENT OF TERRESTRIAL REFRAC-

TION BY RECIPROCAL OBSERVATIONS OF ZENITH
DISTANCES.
Let A and B be two sta-
tions, and let their heights
(ascertained by levelling) be
h and ti. Consider the earth
O
'

as a sphere, and take its

centre. Call the radius r and

the angle v. AOB
Let Z be
the true zenith distance of B
at A, viz., ZAB, and Z' that of
A at B or Z'BA. The dotted
curve shows the path of the
ray of light. A' and B' are the
apparent positions of the sta-
tions,
pig, 4 6.
The co-efficient of refraction is the ratio of the differ-
ence between the observed and real zenith distance at
 Trigonometrical Levelling. 181

either station to the angle v. Thus, if k is the co-efficient

and z z the observed zenith distances, we have k equal to
Z- z Z' z'
or . But these are not always the same.
v v
In the triangle A OB we have
JL
i(Z' + Z)= 9 o- +
2

tan tan J (Z' Z)=

2 h' + 2 r+h
These equations give 71 and Z.

If we substitute for tan the first two terms of its ex-

2
^cu^u^o
pansion in smes, the second equation may be put in the
form

h'h=s tan i
(Z'Z)\i+^~ + i 2 ra )

where s is the length of AB projected on the sea level.

The co-efficient of refraction may also be obtained from
the simultaneous reciprocally-observed zenith distances
of A and B without knowing their heights. Thus :

Z=z +k v, and Z'=z' + k v

: z + z' + z k v i8o +^
z+z' i8o
or i 2 k=
,

v
The mean co-efficient is .0771. For rays crossing the
sea it is .0809, and for rays not crossing it .0750.
The amount of terrestrial refraction is very variable,
and not to be expressed by any single law. In flat, hot
countries where the rays of light have to pass near the
ground and through masses of atmosphere of different
densities the irregularity of the refraction is very great:
so much so that the path of the rays is sometimes
convex to the surface of the earth instead of concave. In
Great Britain the refraction is, as a rule, greatest in the
early mornings towards the middle of the day it de-
;

creases and remains nearly constant for some hours, in-

creasing again towards evening.
 CHAPTER VII

THE USE OF THE PENDULUM IN DETERMINING THE

COMPRESSION OF THE EARTH.

The spheroidal form of the earth causes the force of

gravity to increase from the equator towards the poles,
and this force may be measured at any place by means of
the oscillations of a pendulum.

If we had a heavy particle suspended from a fixed

point by a fine inextensible thread without weight we
should have what is called a simple pendulum. If this
pendulum were allowed to make small oscillations (of not
more than a degree in amplitude) in vacuo, and in a ver-
tical plane, the time of oscillation would be given by the
formula

Where / is the number of seconds, / the length of the

pendulum in feet, and g the force of gravity.

Therefore, taking g as constant, if there were another

pendulum /' feet long and vibrating in t' seconds, we
should have
r
t : t :: \/ I :
\/l'

or, if the time were constant and g changed to g',

VI
 Pendulum Observations. 183

or if / were constant and g variable

If n and ft are the number of oscillations in the time

t and t
1
, then, n' : n::t-t'
:: |/Z :
\/l'
2
t
From (2) wehave, g'^z^g
'2
n
= T~2^ (3)

To find the value of g we can either ascertain by

measurement the length of a pendulum that makes a cer-
tain number of oscillations in a given time, or we can use
a pendulum of invariable length and find g from equation 1

(3). Both methods have been used, but the last is the
easiest in practice.

A simple pendulum as described above is, of course, an

imaginary quantity, and all pendulums actually used are

what are called "compound" pendulums. But it is pos-
sible to calculate the length / of a simple pendulum that
would oscillate in the same time as the compound one,
by finding the position of the "centre of oscillation ;"
that is, of the point which moves in the same manner as
would the pendulum if its whole mass were collected at
that point, thus constituting a simple pendulum. The
centres of oscillation and suspension are interchangeable,
and if a pendulum is suspended from the former, the
latter becomes the new centre of oscillation.

The compression of the earth is calculated thus : If c

is the compression, tp the latitude of a station, g the force

of gravity at the equator, g' that at the station, and m

the ratio of the centrifugal force at the equator to g, we
have, by the formula known as Clairaut's Theorem,
g'=g [i+(fw-c) sin 2 <p\
2
n
and, since g'=-= r- g, if n is the number of oscillations in
n
184 Pendulum Observations.

a given time at the equator and n 1 the number^ at the

station,
n' 2 =-n 2 [i+(| m c) sin 2 <p) (4)

Also,
j-
g
:. if we
take the lengths of the seconds pendulums in-
stead of the number of their oscillations, we have
/.'=/ [i + (f mc) sin 2 <p] (5)
I
being the length of the pendulum at the equator, m
being known, and n n\ or / /',
being found by experiment,
we at once get the value of c from equation (4) or (5).

Borda's pendulum, which was used by the French

astronomers to find the length of the second's pendulum
(that is, a oscillating in a single second) at dif-
pendulum
ferent stations, consisted of a sphere of platinum sus-

pended by a fine wire, attached to the upper end of

which was a knife edge of steel resting on a level agate

plane. The length of the simple pendulum corresponding

to Borda's was obtained by measurement and calculation.
In 1818 Captain Kater determined the length of the
seconds pendulum in London (39.13929 inches) by means
of a pendulum which had two knife edges facing each
other one for the centre of suspension, the other at the
centre of oscillation so that, provided the two knife
edges were at the correct distance apart, they could be
used indifferently as points of suspension the pendulum ;

being, of course, inverted in the two positions. The

pendulum was made to swing equally from either point of
suspension by adjusting a sliding weight. The distance
between the two edges gave the length of the simple pen-
dulum.
The advantage of such a pendulum is that it contains
two in one, and that any injury to the instrument is de-
tected by its giving different results when swung in the
two positions. This pendulum was afterwards super-
seded by another of similar principle, in which, instead of
 Pendulum Observations. 18$
1 -
using a sliding weight, one end of the bar of which it
consisted was filed away until the vibrations in the two
positions were synchronous. In using the pendulum it is

swung pendulum of an astronomical clock,

in front of the
the exact rate of whichis known. By means of certain
contrivances the number of vibrations made by the two
pendulums in a given time can be compared exactly, and
the number made by the clock being known that of the
experimental pendulum is obtained. Certain corrections
have to be applied. One for changes in the thermometer,
which lengthen or shorten the pendulum a second for :

changes in barometric pressure, which by altering the

floatation effect of the atmosphere on the instrument,
affect the action of gravity on it a third for height of
;

station above the sea level, which also affects the force of

gravity, the latter diminishing with the square of the

distance from the centre of the earth and a fourth for
;

the amplitude of the arc through which the pendulum

swings, which, in theory, should be indefinitely small.
The number of pendulum oscillations in a given time
has been observed at a vast number of stations in various
parts of the world, and in latitudes from the equator to
nearly 8o. The most extensive series of observations
was one brought to a close in India, the pendulums
lately
used in which had been previously tested at Kew. The
general of all the pendulum experiments gives
results
about 292 293 as the ratio of the earth's axes, which is
:

the same as that deduced from measurements of meri-

dianal arcs.
 nv>-j/*,-i~<4>(^

Oliver, J R
A course of practical
astronomy

Physical &
Applied So,

PLEASE DO NOT REMOVE

CARDS OR SLIPS FROM THIS POCKET

UNIVERSITY OF TORONTO LIBRARY